THE EFFECTS OF DIGITAL CONTENT ON CUSTOMER ENGAGEMENT ACROSS DIFFERENT TRAVEL PURPOSES By Jung Hee Yu A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Sustainable Tourism and Protected Area Management—Doctor of Philosophy 2021 ABSTRACT THE EFFECTS OF DIGITAL CONTENT ON CUSTOMER ENGAGEMENT ACROSS DIFFERENT TRAVEL PURPOSES By Jung Hee Yu This study investigates how message appeal initiates customer engagement when different trip purposes or hotel attributes are presented in a tourism and hospitality context. This study aims to understand how engagement toward a social media post influences behavioral intention toward a hotel as a common tourism product. Over the last two decades, researchers actively explored the concept of customer engagement. This study pioneers different approaches to the body of existing engagement literature in various aspects: 1) experimental research design, 2) multi-dimensions of customer engagement, 3) engagement toward a social media post, and 4) engagement in the initial exposure to the brand through social media. This study investigates how travel purposes and hotel attributes moderate the effect of appeal on engagement. Finally, additional testing of the existence of a positive relationship between engagement and business performance is added. An experimental research design with two independent studies was executed. Study 1 (n=254) adopted a 2 (message appeal: informational vs. emotional) x 2 (trip purpose: business vs. leisure) full factorial between-subjects design. Study 2 (n=265) employed a 2 (message appeal: informational vs. emotional) x 2 (information topic: core vs. supporting) full factorial between-subjects design. Participants were recruited from the Amazon Mechanical Turk online survey system and directed to the online survey platform Qualtrics. ANCOVA analysis analysis was executed to test the relationship between levels of engagement and behavioral intention. This study’s findings indicate that when travel purpose was studied as a moderator of message appeal on engagement, informational appeal generates a significantly higher page engagement and post engagement than emotional appeal in the leisure travel group. When the message topic was tested as a moderator of message appeal on engagement, emotional appeal elicits a higher engagement level than informational appeal. This study reveals that engagement's psychological aspect is associated with search intention, word-of-mouth intention, and purchase intention. Regarding behavioral aspects of engagement, post engagement alone is not significantly associated with behavioral intention. In contrast, page engagement is associated with word-of-mouth intention. Finally, practical implications are suggested for industry practitioners to optimize social media content. Moreover, travel industry stakeholders should excogitate that leisure travelers engage more with an informational message than with an emotional message when they have limited information about a hotel. iii Copyright by JUNG HEE YU 2021 To my late father-in-law, Sang Hwan Lee v ACKNOWLEDGEMENTS This journey has become longer than we all expected with personal struggles. I received tremendous support from so many people, and without the support, I could not be here. First and the most, I appreciate the tremendous support and patience from my advisors, Prof. Christine Vogt, and Prof. JaeMin Cha, who were dedicated to helping me complete the Ph. D. journey in every aspect, personal and professional. Even when I wanted to give up, they never gave me up. They patiently waited for me, encouraged me to regain myself, supported me academically, emotionally, and financially, guided and helped me move forward. I will remember Prof. Vogt’s devotion toward her students, and I aimed to follow her path. Prof. Cha was devoted to helping me to complete this journey with support and patience. She prayed for me and shared spiritual messages, which lifted me from the bottom of the tunnel. I owe many thanks to other committee members. Prof. Lu Zhang kindly guided me to design, execute, and analyze the experimental research design. I missed our weekly in-person meetings after the pandemic. Prof. Borchgrevink shared his thoughtful insight on theories to advance the research. Prof. McCole was always willing to help, and I could feel that he cared about the students. He also reviewed and edited survey questionnaires. I appreciate their patience and kind support. I want to express special thanks to Prof. Raymond Schmidgall from The School of Hospitality Business (HB), my mentor since 2007. Following his example, I grew up as an vi educator with ethics, discipline, trust, and a warm heart. I was so lucky and happy to be called Prof. Schmidgall’s graduate assistant. Three special faculty members from Community Sustainability (CSUS) helped me overcome the personal barriers I encountered and keep moving forward. Prof. Gail Vander Stoep, Prof. Kimberly Chung, and Dr. Frances Kaneene, thank you so much. Jamie Lyon and Sue Chatterley from HB and CSUS were two angels for me in the departments. Mike and Lucy in the Natural Resources Building ensured us to study until late at night. We felt safe with their presence in the building. My MSU Korean sisters, Profs. Ju Hyung Han, Eun Jeong Noh, Jenni Lee, and Won Kyung Kim are the ones whom I can ask for academic and personal advice anytime. They shared their ideas on the research, proofread the manuscript, and gave me feedback and suggestions to advance this dissertation. Most of all, they assured me that I could complete this journey. CSUS sisters, Prof. Crystal Miller-Eustice (and her family), Dr. Udita Sanga (and her parents), Angela Manjichi (and her family), Becky Bennett, and Natalia Ocampo Díaz, HB sisters, Profs. Miran Kim and Praneet Randhawa Gandhi, you were my family away from home. Thank you so much. I also appreciate all the support from my CSUS and HB friends. My old friends from Korea, Jungmin Lee, Mi Yeon Ahn, Se Eun Lee, Jieun Kim, were always by my side; with their support, I could come this far. I want to express thanks to the MSU Korean Buddhist Organization and Lansing Area Mindfulness Community Meditation groups, and Zen master Hye Bong. Thanks to them, I could grow spiritually during this journey. vii Finally, I owe thanks to my family, Taek Jun Yu, Mi Ja Han, Yu Soon Chang, Sung Gun Yu, Jung Lee, Yina Yu, and Chae Gwang Lee. Their sacrifice made me reach my goal and be here. My husband, Chae Gwang Lee, was my light at the end of the tunnel. Whenever I stumbled and felt no energy to keep going, I could feel the light twinkle and finally come out of the tunnel. As I complete this journey, I wish I can be his light, and we as a team can be a light to others. viii TABLE OF CONTENTS LIST OF TABLES ................................................................................................................ xiii LIST OF FIGURES .............................................................................................................. xiv CHAPTER 1 ............................................................................................................................. 1 INTRODUCTION ................................................................................................................ 1 Statement of the Problem & Main Research Questions ............................................. 3 Statement of Purpose ..................................................................................................... 4 Theoretical Purpose ........................................................................................................ 4 Conceptual Framework ................................................................................................. 6 Delimitations ................................................................................................................... 6 Limitations ...................................................................................................................... 8 Definition of Terms ......................................................................................................... 9 CHAPTER 2 ........................................................................................................................... 11 THEORETICAL BACKGROUND .................................................................................. 11 Literature Review/Overview ....................................................................................... 11 Theoretical Frameworks .............................................................................................. 12 Customer Engagement ................................................................................................. 14 Customer engagement toward brand-related content ......................................... 15 Engagement at the initial exposure to the brand .................................................. 16 Multiple dimensions of engagement ....................................................................... 17 Appeal ............................................................................................................................ 19 Cognitively-affectively based attitude .................................................................... 19 Informational versus emotional appeal ................................................................. 20 Travel Purpose .............................................................................................................. 21 Leisure versus business travel ................................................................................ 21 Hotel Attributes ............................................................................................................ 23 Message topic on hotel attributes (core vs. supporting) ....................................... 23 Behavioral Intention ..................................................................................................... 25 Customer engagement as an antecedent of behavior ........................................... 25 CHAPTER 3 ........................................................................................................................... 28 METHODOLOGY ............................................................................................................. 28 STUDY 1 ........................................................................................................................ 28 Stimuli Development .................................................................................................... 29 A pretest of stimuli ................................................................................................... 29 Travel scenario (business vs. leisure) ..................................................................... 30 Appeal manipulation (informational vs. emotional)............................................. 31 Procedures (Pilot & Main Study) ................................................................................ 31 Manipulation checks ................................................................................................ 32 Dependent variables ................................................................................................ 33 ix Study Participants and Data Cleaning ....................................................................... 35 STUDY 2 ........................................................................................................................ 36 Stimuli Development .................................................................................................... 36 Core topic - bedroom information ......................................................................... 36 Supporting topic - concierge service information ................................................. 37 Procedures (Pilot & Main Study) ................................................................................ 38 Manipulation checks ................................................................................................ 38 CHAPTER 4 ........................................................................................................................... 41 RESULTS ............................................................................................................................ 41 STUDY 1 ........................................................................................................................ 41 Research design ............................................................................................................ 41 Manipulation checks .................................................................................................... 42 Step I: The interaction effect of trip purpose and message appeal on engagement 42 Dependent variables ................................................................................................ 42 Interest and enjoyment ............................................................................................ 46 Page engagement ...................................................................................................... 48 Post engagement ....................................................................................................... 51 Step II: Engagement as a predictor of behavioral intention .................................... 53 Dependent variables ................................................................................................ 53 Search intention ....................................................................................................... 55 Word-of-mouth intention ........................................................................................ 55 Purchase intention ................................................................................................... 56 STUDY 2 ........................................................................................................................ 56 Research design ............................................................................................................ 57 Manipulation checks .................................................................................................... 57 Step I: The interaction effect of message topic and message appeal on engagement ........................................................................................................................................ 58 Dependent variables ................................................................................................ 58 Interest and enjoyment ............................................................................................ 61 Page engagement ...................................................................................................... 63 Post engagement ....................................................................................................... 66 SUMMARY OF RESULTS ......................................................................................... 69 CHAPTER 5 ........................................................................................................................... 72 DISCUSSION ...................................................................................................................... 72 Summary of the Study .................................................................................................. 72 Theoretical Contributions ............................................................................................ 79 Managerial Implications .............................................................................................. 83 Limitations and Future Studies ................................................................................... 86 APPENDICES ........................................................................................................................ 90 Appendix A: Two Instagram Stimuli for Study 1 ..................................................... 91 Appendix B: Four Instagram Stimuli for Study 2 ..................................................... 93 x REFERENCES ....................................................................................................................... 97 xi LIST OF TABLES Table 3. 1. Constructs and measure ......................................................................................... 34 Table 4. 1. Descriptive means of engagement – Study 1 .......................................................... 43 Table 4. 2. Correlations, Means and Standard Deviations of Engagement– Study 1.............. 44 Table 4. 3. Means and ANCOVA results: The impacts of trip and appeal on interest and enjoyment ................................................................................................................................. 47 Table 4. 4. ANOVA results: The impacts of trip and appeal on interest and enjoyment ......... 48 Table 4. 5. Means and ANCOVA results: The impacts of trip and appeal on page engagement .................................................................................................................................................. 50 Table 4. 6. ANOVA results: The impacts of trip and appeal on page engagement ................. 51 Table 4. 7. Means and ANCOVA results: The impacts of trip and appeal on post engagement .................................................................................................................................................. 52 Table 4. 8. ANOVA results: The impacts of trip and appeal on post engagement................... 53 Table 4. 9. Descriptive Means of Behavioral Intention – Study 1 ........................................... 54 Table 4. 10. Correlations, Means and Standard Deviations of Behavioral Intention ............. 55 Table 4. 11. Descriptive Means of Engagement – Study 2....................................................... 59 Table 4. 12. Correlations, Means and Standard Deviations of Engagement– Study 2 ........... 59 Table 4. 13. Means and ANCOVA results: The impacts of topic and appeal on interest and enjoyment ................................................................................................................................. 62 Table 4. 14. ANOVA results: The impacts of topic and appeal on interest and enjoyment ..... 63 Table 4. 15. Means and ANCOVA results: The impacts of topic and appeal on page engagement .............................................................................................................................. 65 Table 4. 16. ANOVA results: The impacts of topic and appeal on page engagement ............. 66 Table 4. 17. Means and ANCOVA results: The impacts of topic and appeal on post engagement .............................................................................................................................. 67 xii Table 4. 18. ANOVA results: The impacts of topic and appeal on post engagement .............. 68 Table 4. 19. Univariate F-values for engagement (interest and enjoyment, page engagement, post engagement) ..................................................................................................................... 69 Table 4. 20. Summary of the results of hypothesis testing ....................................................... 70 xiii LIST OF FIGURES Figure 4. 1. Conceptual map– Study 1 .................................................................................... 42 Figure 4. 2. The interaction effect of trip and appeal on interest and enjoyment ................... 47 Figure 4. 3. The interaction effect of trip and appeal on page engagement ............................ 50 Figure 4. 4. The interaction effect of trip and appeal on post engagement ............................. 53 Figure 4. 5. Conceptual map – Study 2 .................................................................................... 57 Figure 4. 6. The effect of topic and appeal on interest and enjoyment .................................... 63 Figure 4. 7. The effect of topic and appeal on page engagement ............................................ 65 Figure 4. 8. The interaction effect of topic and appeal on post engagement ........................... 68 Figure 5. 1. Significant effects of informational appeal on engagement among leisure travelers - Study 1 .................................................................................................................... 77 Figure 5.2. A significant relationship between interest and enjoyment on search intention, word-of-mouth intention, and buy intention - Study 1 ............................................................. 78 Figure 5. 3. A significant relationship between interest and enjoyment and page engagement on word-of-mouth intention - Study 1 ...................................................................................... 78 Figure A. 1. Informational Appeal Message ............................................................................ 91 Figure A. 2. Emotional Appeal Message ................................................................................. 92 Figure B. 1. Informational Appeal × Core Service Message .................................................. 93 Figure B. 2. Emotional Appeal × Core Service Message ........................................................ 94 Figure B. 3. Informational Appeal × Supporting Service Message ......................................... 95 Figure B. 4. Emotional Appeal × Supporting Service Message .............................................. 96 xiv CHAPTER 1 INTRODUCTION The internet and smartphone technology have evolved, and more platforms are now available; thus, consumers have more options to search for information. For example, before Web 2.0, product information on the internet was mainly found on each product’s official website, and the content was limited to official information controlled by the brand. Now social media has become an important information source. Among tourists, social media plays an even more critical role. Statista (2021) asserted that social media is the most critical player in the travel industry for both tourists and companies. According to their report, approximately 22 percent of U.S. travelers who participated in their study reported that they used social media as a source of inspiration for domestic trips. They also reported National Geographic Travel as the most-followed travel influencer on Instagram, with 39 million followers as of May 2020. Among popular user-generated content platforms, such as Instagram, Facebook, YouTube, and Twitter, Instagram has become the most popular among the young generation, and in planning their travel, they often refer to Instagram (Arefieva, Egger, & Yu, 2021; Filieri, Yen, & Yu, 2021). The content posted on a company’s social media channels is not limited to official information. In digital content marketing, information, words, images, and graphics deliver the story to its target readers, based on readers' information needs, to capture or maintain their attention (Hollebeek & Macky, 2019; Hollimand & Rowles 2014). Digital content marketing has been recognized as a helpful tool, with consumers becoming less influenced by traditional marketing communication (Hollebeek & Macky, 2019). Content components are accentuated as critical marketing components in digital content marketing (Hollebeek & Macky, 2019). As 1 a result, companies have a great deal more leeway in terms of content and valence (e.g., positive/negative) as they create messages for their consumer. Consumer behaviors have expanded as consumers frequently engage with brands on social media and express their feelings for messages about a brand. Social media communications are actively studied by several researchers in different disciplines, including marketing (de Oliveira, Ladeira, Pinto, Herter, Sampaio, & Babin, 2020; Fernandes & Castro, 2020; Hollebeek, Srivastava, & Chen, 2019; Hussein, Hassan, & Ashley, 2020; Kumar, Rajan, Gupta, & Pozza, 2019; Pezzuti, Leonhardt, & Warren, 2020; Shahbaznezhad, Dolan, & Rashidirad, 2021; Syrdal & Briggs; 2018), advertising (Levy & Gvili, 2020; Schivinski, Christodoulides, & Dabrowski, 2016), business/management (Cao, Meadows, Wong, & Xia, 2021; Gutiérrez-Cillán, Camarero-Izquierdo, & San José-Cabezudo, 2017; Harrigan, Evers, Miles, & Daly, 2018; Obilo, Chefor, & Saleh, 2020; Simon & Tossan, 2018), hospitality/tourism (Ferrer-Rosell, Martin-Fuentes, & Marine-Roig, 2020; Filieri, Yen et al., 2021; Gil-Soto, Armas-Cruz, Morini-Marrero, & Ramos-Henríquez, 2019; Tussyaadiah, Kausar, & Soaesilo, 2015; Zhang, Kuo, & McCall, 2019), and information technology (Ferrer-Rosell et al., 2020; Fang & Prybutok, 2018; Oliveira, Huertas & Lin, 2016), and political science (Lappas, Triantafillidou, & Kani, 2021) Information plays a vital role in decision-making processes, and researchers have studied various aspects of consumer information search behavior, such as tourist information search strategies (Fodness & Murray, 1999), information search using social media (Chung Koo, 2015), and information search using smartphones (Ho, Lin, Yuan, Chen, & Alvares, 2016). 2 Consumer engagement behavior is well suited to the context of digital content marketing. The main goal of digital content marketing is to build a long-term relationship with current and prospective consumers by sharing various content that might not be directly related to sales. However, the ultimate goal of marketing is to sell products/services based on the relationship built. Hence, it is vital to measure purchase intention or actual purchase behaviors. Statement of the Problem & Main Research Questions This research study investigates consumer behavior during decision-making. This study examined the relationship between social media messages and consumer engagement in a quasi-experimental study design focused on a hotel reservation situation among prospective consumers. There are three central tenets in the dissertation. The first tenet investigates the impact of message characteristics on readers based on dual information processing models developed from information-processing theories (i.e., central vs. peripheral, cognitive vs. affective). The second tenet investigates whether digital content with information about a hotel can initiate some level of consumer engagement with different motivations (i.e., functional and hedonic). The third tenet investigates whether social media digital content is an object of consumer engagement and if it is possible to observe engagement from initial exposure to the content. The dissertation's primary research question is “How do message characteristics of a post on a hotel’s Instagram page initiate consumer engagement with the hotel and ultimately influence behavioral intentions toward the hotel?” Message characteristics of social media content include information topic/subject (central vs. peripheral) and message content (cognitive vs. affective). 3 Statement of Purpose This dissertation contributes to the body of knowledge regarding empirical testing of social media content and its influence on consumer engagement in several ways. First, this study contributes methodologically by conducting experiments in investigating consumer engagement. Second, this study has a theory-empirical contribution. It attempts to investigate how a message can or cannot create consumer engagement before the physical interaction happens between the brand and customer. Finally, this study offers practical applications to the tourism and hospitality industry by providing them with scientific evidence on cognitive processes associated with social media messages in the context of digital content marketing to further understand how to maximize the use of social media communication in their business. Theoretical Purpose For the last decade, researchers have paid attention to consumer engagement from various perspectives. They have provided evidence that consumer engagement is a predictor of successful business, especially in the long term. Until recently, researchers studied consumer engagement, mainly among current and/or repeat customers, based on the previous brand interaction (mostly physical interaction). Some researchers assume that experience with a brand is a prerequisite for consumer engagement. Through empirical testing, Mollen and Wilson (2010) found that consumer engagement requires “the emergence of the individual’s perceived experiential value and second instrumental value obtained from specific brand interaction.” Bowden (2009) suggested the possibility of consumer engagement among prospective consumers. Bowden (2009) also investigated different consumer engagement processes for new versus repeat consumers of a specific service field (i.e., restaurant dining) and compared the engagement level of new customers to existing customers. The Marketing 4 Science Institute (2010, cited in Brodie, Ilic, Juric, & Hollebeek, 2013) proposed that consumer engagement can exist from pre-purchase to post-purchase contexts. Further, researchers (Bowden, 2009; Brodie, Hollebeek, Jurić, & Ilić, 2011; Vohra & Bhardwaj, 2019) examined different levels and forms of consumer engagement among different customers groups (e.g., current vs. prospective customers). Recently, researchers investigated consumer engagement among perspective consumers (Hollebeek & Macky, 2019; Vohra & Bhardwaj, 2019). Their studies provide a rationale for conducting the experiments among potential consumers who have never used a product/service and adopt a hypothetically created hotel brand to investigate consumer engagement in the initial stage of consumer engagement. Researchers (Bowden, 2009; Vohra & Bhardwaj, 2019) identified that consumers' cyclical processes might differ between new and existing consumers. They asserted that consumer engagement has different stages and, depending on the stage, the level of consumer engagement and the antecedents of consumer engagement vary. Vohra and Bhardwaj (2019) observed engaged consumers who were members of a brand community for varying periods. They reported that no differences in engagement level were detected among members who had been in the community for more than one year. On the other hand, they observed a significant level of differences in engagement among new members whose membership is shorter than one year. Previous research mainly studied consumer engagement in the context of the existing brand community. This research investigates how a message can or cannot create consumer engagement even before the physical interaction happens between the brand and customer in the context of digital content marketing. 5 Beyond theoretical applications, an applied purpose exists for the industry. Marketers often test different ways to manipulate content on social media. The findings from this study provide marketers with evidence on how social media messages perform in the context of digital content marketing. Conceptual Framework Several theories were considered to guide this research study with a focus on attitudes and intentions to act. The theory of planned behavior (TPB; Ajzen, 1991) has been widely adopted and tested to predict behaviors (Ajzen, 2011). This study extends TBP by empirically testing consumer engagement as a distinct concept from attitude as an antecedent of behavioral intention. It also integrates the elaboration likelihood model and a cognitively- affectively based attitude (cognitive vs. affective) in manipulating messages and scenarios. The scope of this study is delimited as follows: Delimitations 1. In this study, the information was delimited to a brand (e.g., the hotel), and information attributes were studied. The scope of this study was to examine consumer engagement toward the brand-related content on social media instead of the brand itself. In identifying antecedents of engagement behavior, Fang et al. (2018) observed that information source attributes showed the most significant impact on user’s engagement behavior followed by information attributes. 2. The valence was controlled to a positive valence only. The valence of a message on social media has been widely studied, and negative reviews have been revealed to have a substantial impact on consumer’s perceptions (Lee, Jeong, & Lee, 2017). The proposed research was designed from the perspective of marketers. The message was 6 assumed to have been created by the hotel, and marketers are likely to emphasize the positive aspects of their product/service. 3. This research only investigated new consumers and not existing consumers. In consumer engagement literature, research supports that consumer engagement is different among prospective vs. repeat consumers. For example, an affective message plays a more important role than a cognitive message among repeat consumers. On the other hand, message relevance, which is one of the characteristics of a cognitive message, is more critical among prospective consumers. 4. This study used a fictional product (i.e., hotel), and participants did not have either any previous information or experience with the product. In the engagement literature, studies have investigated engagement with known brands. In developing consumer engagement measurement in tourism brands, So, King, and Sparks (2014) surveyed consumers who had previously interacted with the brand. 5. This study tested aspects of consumer engagement observed only in the initial stage of consumer engagement, such as attention, interest/enjoyment, and initial intentions, to measure the psychological state of mind caused by social media content (Syrdal & Briggs, 2018). In the previous literature, some essential components of consumer engagements were identified based on actively engaged consumers (e.g., mutual exchange of information, support among members; Vohra & Bhardwaj, 2019). Aspects of consumer engagement observed among actively engaged consumers are not applied to this study. 7 The study is designed with an awareness of limitations as described below: Limitations 1. This research design contains weaknesses related to any experimental design and weaknesses from an artificial situation. In actual Instagram pages, consumers are exposed to multiple posts, and they can read other consumers’ reactions toward posts or the brand page through the number of “likes” or comments. In this study, only one post by the hotel was exposed to the participants, and there was no function for likes or comments. 2. This research adopted a quasi-experimental design having no true control group. In a true experimental design, the subjects in the control group are assumed to be free from the effect of the treatment, and a cause-and-effect relationship between the treatment and dependent variables is more robust than in a quasi-experimental design. In this research, subjects were randomly assigned to one of the four groups for each experiment, thus making the design quasi-experimental and containing the limitation of a quasi- experimental design. 3. Variables beyond those being manipulated may influence dependent variables. An individual’s intrinsic characteristics play an essential role in making a purchase decision or engaging with brands. This study measured personality, online activity, travel experience, and hotel reservation experience to identify confounding effects and control latent variables related to intrinsic characteristics. However, it is still impossible to control all of the latent variables. Being aware of this weakness in the experimental design, the study adopted random assignment of subjects. However, to have a successful analysis, the homogeneity among assigned groups was confirmed. 8 4. Control factors have a limitation in an experiment with participants who are recruited through an online panel. For example, it is hard to monitor the participants during the experiment, aside from monitoring the amount of time participants spent on the experiment, which raises the possibility of a validity issue. 5. Actual hotel purchase was not measured. Instead, this research measured the behavioral intentions of the participants. Some researchers question the gap between behavioral intention and actual behavior. Even though many researchers have explained the gap, there is not any guarantee that a gap between behavioral intention and actual behavior does not exist. Definition of Terms The definitions of important concepts of this study are presented here: Behavioral intention: “The agent's subjective probability that he or she will perform the behavior” (adopted from Ajzen & Fishbein, 1980). Business trips: “Trips undertaken for purposes related to work (Davidson, 1994; cited in Radojevic, Stanisic, Stanic, & Davidson, 2018). Core information: “information providing a rational response using criteria such as information quality” (adapted from Petty, Cacioppo, & Schumann 1983). Customer engagement: “A psychological state of mind experienced when consuming social media content in which an individual is highly absorbed in the content and experiences a sense of excitement” (adopted from Syrdal & Briggs, 2018). Emotional appeal: “a brand attribute information available to the consumer indicates a liking or pleasurable attitude toward the brand” (adapted from Oliver, 1999). 9 Functional motive: “The need for facilitating the choice and consumption of a product or experience” (adopted from Vogt, 1993; citing Bettman, 1979). Hedonic motive: “The need of serving the psychological, pleasure, or entertaining experiences of consuming a product or information” (adopted from Vogt, 1993; citing Holbrook & Hirschman, 1982). Hotel core attribute: In a hotel context, the core feature represents the hotel's primary reason for being in the market. It comprises the hotel's fundamental competency in creating value with and for the customer (Ferguson, Paulin, Pigeassou, & Gauduchon, 1999). Examples of core features include the cleanness and comfortability of hotel rooms. Hotel supporting attribute: In a hotel context, the supporting feature supports or facilitates the delivery of the core offering (Browning, So, & Sparks, 2013). An example of service features is customer services such as concierge and front desk. Informational appeal: “a brand attribute information available to the consumer indicates that one brand is preferable to its alternatives” (adopted from Oliver, 1999). Leisure trips: “Trips undertaken for pleasure, with their motivations including rest and relaxation; spending time with friends and family; meeting new people; shopping; attending sports events; visiting historical and cultural sites; or experiencing places perceived to be exotic, romantic, or having good scenery or nice weather” (Lee, Hsu, Han, & Kim, 2010, Murphy et al., 2007; cited in Radojevic et al., 2018). Supporting information: “simple rules or information shortcuts such as brand image and source attractiveness that consumers use to assess a recommendation rather than evaluating the quality of the arguments used by a source” (adopted from Petty & Cacioppo 1986). 10 CHAPTER 2 THEORETICAL BACKGROUND Literature Review/Overview This study examines consumer behavior and marketing in the context of tourism. One overarching question for this study is, “How does marketing in social media change a consumer’s behavior in information processing?” This question is answered with a focus on customer engagement behavior using social media. Businesses, particularly in the travel and tourism industry, increasingly rely on social media as a marketing tool by posting messages in different forms (e.g., text, picture, video). Consumers react to social media marketing, exhibiting a relatively new behavior in which they express how they feel about the message or about the business entity itself, which can, in turn, spread that information and influence others (So et al., 2014). Researchers have been exploring this behavior under the concept of customer engagement. This study is designed to expand on that body of research and empirically test customer engagement to predict intention to search for more information, share and influence others, and make a purchase. Researchers may investigate a message’s topic, attributes, valence, tone, source, or more (Andreu, Casado-Díaz, & Mattila, 2015; Fang & Prybutok, 2018; Gutiérrez-Cillán et al., 2017; Hwang, Choi, & Mattila, 2018; Lee et al., 2017). This study focuses on a message’s effectiveness, based on dual information processing models, examining central versus peripheral topics and cognitive versus affective content. Message strategy is examined as a predictor/antecedent of consumer behavior in the travel and tourism industry in the context of a consumer’s reason for travel. In the proposed experimental design, the fundamental scaffold is from extended models of the theory of planned behavior (TPB; Ajzen, 1991). As an 11 essential predictor of behavioral intention, customer engagement toward a message is examined in a social media context. Theoretical Frameworks TPB states that attitude, subjective norms, and perceived behavioral control can be used to successfully predict intentions and behaviors (Ajzen, 2011). Based on TPB, this research measures customer engagement caused by manipulations in message appeal (informational vs. emotional), message topic (core vs. supporting), and purpose of travel (hedonic vs. functional). The behavioral intention was tested as an endogenous variable. Researchers (Leung & Bai, 2013; Leung & Jiang, 2018; Xiang & Gretzel, 2010) identified that social media played a significant role in the travel industry, such as leading to changes in travel information searches, travel planning, and decision-making behaviors, and have studied consumer behaviors in social media platforms. Leung and Jiang (2018) applied TPB and tested the models to predict intentions and behaviors in the context of hospitality, studying consumer behavior in the social media platforms of hospitality business entities. In their study, Leung and Jiang (2018) examined the consumer attitudes induced by an Instagram page, using them as a critical indicator in measuring Instagram marketing outcomes. While they examined an extended TPB model in the context of Instagram marketing, they did not differentiate the concept of customer engagement behavior from consumer attitude. Over the last two decades, however, the concept of customer engagement has been studied as a unique entity, so testing consumer engagement in the extended model of TPB is valuable. In developing a model to test the antecedents of customer engagement, this study adapted dual-process models in manipulating messages and scenarios. The elaboration likelihood model (Petty & Cacioppo, 1986) and a cognitively-affectively based attitude 12 (Zajonc,1980) were adapted to create message stimuli for message appeal (informational and emotional), travel purpose (leisure vs. business), and message topic (core vs. supporting). Jun (2009) applied and tested information-processing models and identified three dual-process models (an elaboration likelihood model, a heuristic-systematic model, and a cognitively- affectively based attitude) for studying different information needs in information-processing and found that individuals utilized distinct features of information in different situations. Elaboration likelihood model (central vs. peripheral routes) The elaboration likelihood model (ELM; Petty & Cacioppo, 1986) explains the cognitive processes of how different message characteristics impact information processing, suggesting two distinct routes (central vs. peripheral routes). In ELM, a central route triggers the receiver’s cognitive activity to process information more deeply and critically than a peripheral route (Leong, Hew, Ooi, & Lin, 2019). For example, in the context of advertising messages, functional components that deliver the benefits of a product activate the central route. On the other hand, hedonic components that deliver embedded cues in a message activate the peripheral route. Jun (2009) suggested the terms “effortful mode” and “effortless mode” in explaining ELM. According to Jun (2009), the effortful mode is activated in the central route because information processing requires rationality and logical thinking. For example, individuals in the central route pursue answers to specific questions to aid their decision-making. In the peripheral route, in which the effortless mode is activated, individuals only require simple information cues to make a decision. Morris et al. (2019) highlighted analytical attention and scrutiny in the central route, associating the central route with high cognitive elaboration. 13 They explained the peripheral route as narrative transportation, illustrating the “mind” as becoming active and making “reality” fade in the background. Involvement is a critical concept in the ELM. For example, Rather, Hollebeek, and Rasoolimanesh (2021) studied the effect of appel on customer engagement and treated involvement as a moderator. In extending their study, this study examined the effect of message appeal on customer engagement when the different level of involvement (high versus low) was imposed by dichotomous travel purposes (business versus leisure) in the scenario and hotel attributes (core versus supporting) in the social media message. Customer Engagement With Web. 2.0 and the advent of social media, consumers now use those platforms to share their feelings and reviews. Over the last two decades, these behaviors have been observed and identified by researchers as one aspect of customer engagement. Business entities, including tourism brands such as Marriott and Cathay Pacific, use social media to engage their consumers by providing brand-related social media platforms and enabling their consumers to share information, opinions, and experiences with each other (So et al., 2014). The concept of customer engagement has been studied widely as an emerging concept in the marketing literature since 2005 (Brodie et al., 2011; Hollebeek, Glynn, & Brodie, 2014). Kotler, Kartajaya, and Setiawan (2016) emphasized that marketers should convert first-time buyers into loyal advocates using series of customer engagement tactics. More recently, empirical studies 6that go beyond conceptual studies have been conducted in various fields (Mirbagheri & Najmi, 2019). 14 Customer engagement toward brand-related content Despite a great deal of research on engagement, most empirical studies have studied actively engaged/existing consumers. However, conceptually, researchers opened the possibility to observe some aspects of engagement among potential consumers. To apply customer engagement based on the initial contact (via a social media message), this study limits the scope of the study to the object of customer engagement with the brand-related content on social media instead of the brand itself. More recently, researchers (Fang & Prybutok, 2018; Hollebeek & Macky, 2019; Mirbagheri & Najmi, 2019; Schivinski et al., 2016; Syrdal & Briggs, 2018) paid attention to engagement toward content. Syrdal and Briggs (2018) highlighted information as the attraction in customer engagement, not the brand. That is, they argued that the focal point of social media engagement is the content, not the brand. Mirbagheri and Najmi (2019) identified customer engagement as a context‐specific construct and studied customer engagement as a static state of mind that is influenced by post, content, and campaign in various media formats. Fang and Prybutok (2018) asserted that the existing studies of engagement behavior have not studied psychological aspects of engagement behaviors and argued that posting-related attributes also might have a significant impact on users' engagement behavior. Following their evidence, the proposed research assumes that messages can create customer engagement. Syrdal and Briggs (2018) suggested an engagement with social media content (ESMC). They pointed out the uniqueness of ESMC from interactive behaviors observed in social media and argued that engagement is not necessarily a requirement of interaction. They stated that marketing practitioners regarded the characteristics of ESMC as active behaviors 15 (e.g., liking, commenting, sharing). On the other hand, consumers viewed the characteristics of ESMC as a state of mind, such as enjoyment with a high degree of involvement. Engagement at the initial exposure to the brand If customer engagement is limited to interactive behaviors, it is restrictive to observe customer engagement among potential consumers who have never been exposed to a brand. Researchers who focused on customer engagement among current customers mainly examined interactivity related to brand and value co-creation aspects (Brodie et al., 2013; Mollen & Wilson, 2010). Their studies are based on current/existing users with the assumption that the users have previous experience of community participation, either active or passive. Recently, however, more diverse aspects of customer engagement behaviors in different stages of consumer interaction with the brand have been explored. For example, Kumar et al. (2019) investigated customer engagement in the service domain. They suggested market type (emerging vs. developed market) as one of the moderating variables between firm-related factors and customer engagement. In their study, they emphasized that it is crucial to provide cognitive information in an emerging market, where there are a large number of non-users. In studying customer engagement, Hollebeek et al. (2019) identified different patterns between new and existing consumers. New consumers focus on investing primary resources; existing consumers, on the other hand, invest more complex, more in- depth, and broader resources in brand interaction. Rather et al. (2021) investigated distinctive patterns among new and existing customers in the tourism context. They found that first-time visitors to a destination are more cognitively engaged with a destination, while repeat visitors showed high emotional destination engagement. They suggested destination markets to provide practical site-related information to first-time clients. 16 Multiple dimensions of engagement Even though the concept of engagement has been studied over the past two decades, there is a lack of agreement in the overarching conceptual domain of engagement (Rather et al., 2021; Syrdal & Briggs, 2018). Syrdal and Briggs (2018) asserted that “clarifying the meaning of social media engagement is currently a top marketing research priority in academia.” Rather et al. (2021) argued that despite the growing interest in customer engagement among practitioners and researchers in the past decade, definitions of CE are still debatable. Although most studies have measured the dimensions of customer engagement by observing interactive behavior on social media, such as liking, commenting, and sharing (Shahbaznezhad et al., 2021), more researchers (Mirbagheri & Najmi, 2019; Mollen & Wilson, 2010; Schivinski et al., 2016) are studying, engagement level with multiple dimensions. For example, Mollen and Wilson (2010) asserted that “engagement is a discrete construct comprised of cognitive and affective dimensions. Brodie et al. (2013) suggested a multidimensional concept of customer engagement with cognitive, emotional, and/or behavioral dimensions. Schivinski et al. (2016) adopted a multi-level approach in measuring customer engagement and suggested three (multi-level) dimensions of engagement with brand-related social media content as consumption (lower-level), contribution (higher-level), and creation (higher-level). Other researchers agreed that customer engagement is a multidimensional construct composed of cognitive, emotional, and behavioral dimensions (Dessart, Veloutsou, & Morgan-Thomas, 2015; Hollebeek, 2011; Mirbagheri & Najmi, 2019; So et al., 2014; Vivek, Beatty, & Morgan, 2012). Brodie et al. (2013) analyzed the nature of customer engagement focused on consumers’ specific interactive experiences, engagement objects, motivational states, and engagement dimensionality within a brand community. The 17 three aspects of customer engagement (cognitive, emotional, and behavioral) are revealed to interplay and affect one another. From their qualitative analysis of online communities, they found that the emotional aspect of consumers’ online brand engagement is negative versus positive, low- versus high-intensity, and short- versus long-term attitude. They observed emotional feelings as consisting of gratitude, empathy, trust, feeling safe, and a sense of belonging to the group (i.e., social aspect). They identified the cognitive aspect of customer engagement as a value-laden relationship through sharing information and experiences. Finally, customer engagement behavioral dimensions were observed as being the posting behavior indicating members’ participation in the online community. In investigating how critical social media contextual factors influence social media engagement behavior, Cao et al. (2021) viewed social media engagement behavior as having various levels and adopted three dimensions to measure social media engagement behavior (i.e., consumption, contribution, and creation). This study viewed and measured engagement in both psychological and behavioral aspects; and applied findings from recent studies to inform the design. Syrdal and Briggs (2018) defined engagement as a state of mind reflecting the consumers’ perspective based on the findings and suggested that there are two different approaches to study customer engagement: first, as a specific state of mind; and second, as a process of moving into and out of a state. In this study, the psychological aspect of engagement was measured with the view of engagement as a state of mind when consumers are first exposed to the brand-related message. In the behavioral aspect of engagement, this study adopted interactive behavior on social media in two levels: a social media post level and a social media page level. Researchers (Villamediana-Pedrosa, Vila-López, & Küster-Boluda, 2020) view interactive 18 behaviors as engagement based on a relational marketing perspective. In studying customer engagement, Brodie et al. (2013) studied customer engagement based on relationship marketing theory, emphasizing the relational concepts through the interactive experiences between the consumers and the brand and/or other members of the brand community. Schivinski et al. (2016) reported that the interactive nature of social media changed how consumers engage with brands, viewing engagement as expressed by interactive behaviors such as reading, writing, commenting, liking, sharing, etc., on social media. Cognitively-affectively based attitude Appeal A cognitively-affectively based attitude (Zajonc, 1980) presents contrasting concepts of cognitive and affective responses. According to Levenson (2019), the study of cognition and emotion (affect) became prominent in the 1980s, as Zajonc (1980) first argued affect is an independent concept from cognition. Unlike previous research on the cognitive reactions of the 1970s, researchers in the 1980s studied emotion as a separate concept from cognition. They actively investigated the relationship between cognition and emotion (affect) (Levenson, 2019). Zajonc (1984) strongly disagreed with the view that cognitive appraisal was a prerequisite of affective arousal and emphasized the independence of effect from cognition. Zajonc (1980) explained that an affective response is associated with the expression of emotion while a cognitive response requires objective judgment emphasized in law or science, requiring more cognitive effort. He concluded that affection and cognition are under the control of separate and partially independent systems that can influence each other in a variety of ways and that both constitute independent sources of impact in information processing. 19 Other researchers (Levenson, 2019; Poels & Dewitte, 2019) emphasized the influence of emotion on cognition and asserted that emotion has a significant influence on our cognitive processes such as thoughts, judgment, reasoning, consciousness, and communication method. Further, Poels and Dewitte (2019) asserted the vital role of emotion (affect) in information processing and behavior, especially in the context of advertising. In examining advertising- relevant behavioral outcomes, they suggested that a digital media environment has more potential for stimulating emotions. Levenson (2019) found this to be especially true when a peripheral route is activated. Poels and Dewitte (2019) perceived emotion as a part of the affective process and explored behavioral outcomes triggered by emotions in the context of advertising. They understood emotions as separate groups of integral and incidental emotions. In advertising, integral emotions are used to influence consumers by evoking emotions that are deliberately and strategically embedded in the message. Mayer and Tormala (2010) applied a cognitively-affectively based attitude into a “think versus feel” message frame to explain how the recipient is affected by the orientation (cognitive vs. affective) of an advertisement’s message. Chen et al. (2015) adopted affective and cognitive elaborations on attitude formation using branded content on Facebook. Their findings suggested that affective and cognitive elaborations work simultaneously. However, they found that effect has more influence than cognition in consumer attitude formation and suggested that this shapes the decision-making process. Informational versus emotional appeal In customer engagement literature, information is considered to be an essential factor. Brodie et al. (2013) provided an exploratory empirical study that yielded evidence to support the five themes observed in customer engagement. Their analysis shows the customer’s need 20 for information to initiate customer engagement. According to Bowden (2009), a new customer’s engagement is cognitive in nature, with a calculative commitment, whereas repeat customers show a more emotional aspect, with an affective commitment. Based on this finding, he developed an assumption that cognitive and central messages can cause engagement among new customers. However, the degree of a message’s impact on consumers could vary based on their motivation (e.g., hedonic vs. functional). Even though researchers emphasized the effect of affective elaboration in attitude formation, other researchers argued that cognitive elaboration under certain conditions, such as when functional aspects became emphasized (e.g., for the new consumers rather than repeat consumers) with high involvement, cognitive process, acted more strongly. This study proposes the following hypothesis to test the differences among new customers: Hypothesis 1: There will be the main effect of message appeal on engagement exhibited by (a) interest and enjoyment, (b) page engagement, and (c) post engagement. Leisure versus business travel Travel Purpose Researchers (Bi, Liu, Fan, & Zhang, 2020; Galati & Galati, 2019; Kim & Park, 2017; Radojevic et al., 2018; Yavas & Babakus, 2005) identified the two major market segments of the hotel industry as leisure and business. They provided evidence that leisure travelers and business travelers show different patterns in various areas, including hotel selection and evaluation criteria, preference for hotel attributes, and expectations. For example, Kim and Park (2017) asserted that while business travelers considered room quality and a comfortable 21 feeling as hotel reservation priorities, leisure travelers with family weighed more on price and overall atmosphere. Yavas and Babakus (2005) asserted that the different priorities in hotel choice configurations for each group should be reflected in the marketing approach targeting each segment. Jones and Chen (2011) conducted an experimental study to explore information search behavior for hotel selection. They provided evidence that business travelers and leisure travelers displayed different patterns in information search. They confirmed that leisure travelers are highly involved in hotel reservations, and when they reserve a hotel room, they require complex decision-making models. Literature supported that leisure travel is associated with high involvement and creates a central route in the ELM (Bi et al., 2020; Jones & Chen, 2011; Radojevic et al., 2018). Business travel was assumed to create the peripheral route in information processing. Bi et al. (2020) asserted that leisure travelers are more critical about the hotel than business travelers, as leisure travel expenses are covered by themselves. Further, business travelers often have others make bookings, and decision-making on where to stay is at a corporate level. Radojevicet al. (2018) mentioned that an assumption of leisure travelers being more price- sensitive than business travelers was frequently highlighted. They also addressed that leisure travelers might enjoy their leisure trips more fully than business travelers. Business travelers had limited ability to enjoy their stay at the hotel as their preference for hotel and destination attributes are lightly considered in their travel. In tourism research, there is limited empirical research that tests decision-making and information search processes in both leisure and business contexts. This research is groundbreaking on framing travel purposes, and in an exploratory way, uses a moderator variable approach to assess whether travel purpose is an influential variable. 22 Based on ELM and the assumption based on the literature, the following hypotheses were developed: Hypothesis 2A. Trip purpose moderates the effect of message appeal on the level of engagement. More specifically H2Aa: In the functional (vs. hedonic) trip scenario, the level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will be higher with emotional appeal than with informational appeal. H2Aa(a). In the functional trip scenario, the level of interest and enjoyment with the message will be higher with emotional appeal than with informational appeal. H2Aa(b). In the functional trip scenario, the level of engagement page engagement with the message will be higher with emotional appeal than with informational appeal. H2Aa(c). In the functional trip scenario, the level of engagement (post engagement with the message will be higher with emotional appeal than with informational appeal. H2Ab. In the hedonic (vs. functional) trip scenario, the level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will be higher in informational appeal than in emotional appeal. H2Ab(a). In the hedonic trip scenario, the level of interest and enjoyment with the message will be higher in informational appeal than in emotional appeal. H2Ab(b). In the hedonic trip scenario, the level of page engagement with the message will be higher in informational appeal than in emotional appeal. H2Ab(c). In the hedonic trip scenario, the level of post engagement with the message will be higher in informational appeal than in emotional appeal. Message topic on hotel attributes (core vs. supporting) Hotel Attributes Marketing researchers (Browning et al., 2013; Filieri, Galati, & Raguseo, 2021; Kumar et al., 2019; Morrison, 2002; Yen & Tang, 2019) frame product attributes into two categories – core and supporting (consequential in Morrison, 2002). A core hotel attribute is the hotel's prime reason for being in the market and covers the hotel's essential expertise in 23 creating value with and for the guest (Ferguson et al., 1999); a supporting hotel attribute is supporting features which support or facilitate the delivery of the core offering of a hotel and add value to the service package (Browning et al., 2013). Research on hotel attributes has shown that a room feature is a core service in a hotel and asserted that online reviews on the core service have a greater effect on consumer perceptions of the hotel (Browning et al., 2013). Xiang, Schwartz, Gerdes, and Uysal (2015) identified guest rooms, beds, and bathrooms as examples of a hotel's core product. Yen and Tang (2019) summarized that the tangible aspects of the hotel, including the guest room as core attributes, and failures in delivering satisfying core attributes could lead to complaints from customers. Based on the literature, this study assumed that an Instagram post with core hotel attributes creates a central route. On the other hand, an Instagram post with supporting hotel attributes (i.e., concierge information) was assumed to create the peripheral route in this study's information processing. Based on ELM and the assumption based on the literature, the following hypotheses were developed: Hypothesis 2B. Message topic moderates the effect of message appeal on the level of engagement. More specifically H2B a: With core (vs. supporting) information, the level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will be higher with emotional appeal than with informational appeal. H2B a(a). With core (vs. supporting) information, the level of interest and enjoyment with the message will be higher with emotional appeal than with informational appeal. H2B a(b). With core (vs. supporting) information, the level of engagement page engagement with the message will be higher with emotional appeal than with informational appeal. H2B a(c). With core (vs. supporting) information, the level of engagement (post engagement with the message will be higher with emotional appeal than with informational appeal. 24 H2B b. With supporting (vs. core) information, the level of engagement ((a) interest and enjoyment, (b) page engagement, and c) post engagement) with the message will be higher in informational appeal than in emotional appeal. H2B b(a). With supporting (vs. core) information, the level of interest and enjoyment with the message will be higher in informational appeal than in emotional appeal. H2B b(b). With supporting (vs. core) information, the level of page engagement with the message will be higher in informational appeal than in emotional appeal. H2B b(c). With supporting (vs. core) information, the level of post engagement with the message will be higher in informational appeal than in emotional appeal. Customer engagement as an antecedent of behavior Behavioral Intention From the perspective of marketers, social media interactive behaviors are viewed as an indicator of marketing success. Syrdal and Briggs (2018) identified the benefit of engagement as increased sales, brand loyalty, and brand equity. Schivinski et al. (2016) utilized the engagement scale as an instrument for auditing and tracking the effectiveness of marketing strategies to measure the construct’s effects on outcome variables such as brand extension, purchase intention, and price premium. Brodie et al. (2013) investigated customer engagement in online brand community contexts and observed that customer engagement caused loyalty, commitment, and empowerment in that context. In addition to those prominent concepts, Brodie et al. (2013) added trust, self-brand connections, and emotional brand attachment as a byproduct of customer engagement. Further, Mollen and Wilson (2010) suggested a sequential online engagement model as an antecedent of attitude and behavioral intention. As stated above, Brodie et al. (2013) 25 identified customer engagement as an antecedent of consumer loyalty, satisfaction, consumer empowerment, connection, emotional bonding, trust, and commitment. Moliner, Monferrer- Tirado, and Estrada-Guillén (2018) hypothesized and reveal the positive relations between customer engagement and the firm’s financial performance. Based on the literature, the following hypotheses were developed: Hypothesis 3. The level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will positively influence the behavioral intention (a. search intention, b. word-of-mouth intention, c. purchase intention) H3a. The level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will positively influence the search intention. H3a(a). The level of interest and enjoyment with the message will positively influence the search intention. H3a(b). The level of page engagement with the message will positively influence the search intention. H3a(c). The level of post engagement with the message will positively influence the search intention. H3b. The level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will predict the level of word-of-mouth intention. H3b(a). The level of interest and enjoyment with the message will positively influence the word-of-mouth intention. H3b(b). The level of page engagement with the message will positively influence word-of-mouth intention. H3b(c). The level of post engagement with the message will positively influence the word-of-mouth intention. H3c. The level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will predict the level of purchase intention. H3c(a). The level of interest and enjoyment with the message will predict the level of purchase intention. H3c(b). The level of page engagement with the message will predict the level of purchase intention. H3c(c). The level of post engagement with the message will predict the level of purchase intention. 26 This study was designed to investigate how engagement toward a social media post influences behavioral intention toward the hotel among the new customers. In studying the concept of customer engagement, most studies were focused on engagement toward brand or community among existing customers. In the methodological aspect, content analysis or using the secondary data was predominant. This study shed different perspectives on the body of existing engagement literature in various aspects: 1) experimental research design, 2) multi- dimensions of customer engagement, 3) engagement toward a social media post, and 4) engagement in the initial exposure to the brand through social media post. In investigating the effect of message appeal on engagement, this study applied tourism and hospitality context by testing travel purposes and hotel attributes as moderators between message appeal and engagement. 27 CHAPTER 3 METHODOLOGY This research conducted two separate experiments to test travel purpose (leisure vs. business) and hotel attribute (core vs. supporting) as moderators of appeal on engagement. In the two separate experiments, travel purpose and message topic were studied as moderators of appeal on engagement in each experiment based on the literature. Previous research (Rather et al., 2021) studied customer involvement as a moderating variable of engagement on revisit intention. This research adopted travel purpose (Villamediana-Pedrosa et al., 2020) and hotel attribute (Browning et al., 2013; Dimitriadis & Koritos, 2014; Filieri, Galati et al., 2021; Lovelock & Wirtz, 2011; Slevitch & Oh, 2010; Yen & Tang, 2019) to control for involvement level. The design is considered to be quasi-experiment and not a true experiment. A control group and pre-and-post measures were not features of the design. Study 1 adopted a 2 (message appeal: informational vs. emotional) × 2 (trip purpose: business vs. leisure) full factorial between-subjects design. Study 2 employed a 2 (message appeal: informational vs. emotional) × 2 (information topic: core vs. supporting) between-subjects design, with subjects being randomly assigned to one of the four messages represented by different combinations of the two independent variables. Amazon Mechanical Turk's online survey system (MTurk, https://www.mturk.com) was used to recruit, and participants were directed to Qualtrics, an online survey platform. STUDY 1 Study 1 adopted a 2 (trip purpose: business vs. leisure) x 2 (message appeal: informational vs. emotional) full factorial between-subjects design. This chapter discusses the process by which the stimuli were developed. Two travel scenarios with a different purpose 28 (business vs. leisure) for visiting the hotel were adopted as the stimuli from previous literature. Next, the manipulation of the stimuli (i.e., Instagram posts with a different appeal- informational vs. emotional) is described, including the stimuli development, a pretest, a pilot study, and the main study for Study 1. The detailed procedure of each step is discussed in this chapter. Manipulation checks, measures, study participants, and data cleaning methods are explained. The results of statistical analysis are discussed in the following chapter. A pretest of stimuli Stimuli Development Before administering the experiment, a pretest of stimuli was conducted through Qualtrics, an online survey platform. The pretest was designed to confirm and improve the reality of the travel scenario and two social media posts with different message appeals (informational and emotional). Active social media users, who have "liked" or commented on the researcher's social media page, were recruited. Participants were invited to an online survey link that directed them to the Qualtrics website. Participants were asked to identify flaws in the stimuli and suggest their opinion on improving the stimuli. Twenty-one participants took part in the pretest. Initially, the stimuli were created in the format of a Facebook post. However, three of the participants recommended using the Instagram post. Researchers identified Instagram as the most used social networking platform and as a platform reflecting significant communication trends in promotional activities (Ilich & Hardey, 2020). After seeking an expert’s advice on such an issue, Instagram was chosen to represent the social media platform suitable for this study. An Instagram account (https://www.instagram.com/hotel_zee) was created with profile information stating, "Hypothetically Created Hotel for Study." The 29 messages were posted on Instagram, screen-captured, and deleted. The screen-captured messages were embedded in the survey on Qualtrics for the pilot and main study. Travel scenario (business vs. leisure) Initially, the travel scenario used in the pretest included the hotel's price information, the reviewer's star rating, and the number of reviews in addition to the general travel scenario. However, participants expressed their confusion related to that information. After consulting with an expert and reviewing further experimental studies, the scenario has been adapted from previous studies (Nguyen et al., 2016; Taylor & Kimes, 2010). The neutral travel purpose scenario read as follows: You are going out of town on a trip for a couple of days. You will be staying at a four-star hotel (similar to the Holiday Inn, Four Seasons, Novotel, Radisson Hotel, and Suites) that offers semi high-end amenities such as personalized service, 24-hour room service, and valet parking, a fitness center, and a full-service restaurant. You found Hotel XYZ on the internet that matches your needs. The scenario was manipulated into two versions of business purpose, and leisure and participants were randomly assigned to one of the two travel scenarios. In the functional scenario condition, participants were asked to imagine, "You are going out of town on a business trip for a couple of days." In the hedonic scenario condition, participants were asked to imagine, "You are going out of town on a leisure trip for a couple of days" The neutral travel purpose scenario was adopted for Study 2. 30 Appeal manipulation (informational vs. emotional) Informational appeal messages and emotional appeal messages were adapted from a previous study (Wu et al., 2017). The informational (rational) appeal message was designed to deliver facts directly and objectively (Andreu et al., 2015). For example, in describing the spaciousness of the room, the square feet of the room (i.e., 500 sq. ft. room) was stated. To be consistent with the travel scenario given to the participants, the room size information was extracted from an actual four-star hotel's website. The affective appeal message was designed to evoke feelings and create a positive feeling about the room (Andreu et al., 2015). For example, in describing the spaciousness of the room, the phrase, "bigger rooms than those in a king and queen's palace," was adapted from a previous study (Wu et al., 2017). For Study 1, two messages were created (Appendix A). Other than message appeal (informational or emotional) manipulation, the contents of the information and the picture of the two messages remained consistent across the two conditions. Procedures (Pilot & Main Study) On the first page of the survey on Qualtrics, an IRB (the Institutional Review Board) approved informed consent was provided, and the participants agreed to the consent by clicking the next page button on the website. Then, a screening question on participants' online hotel room reservation experience was asked with a seven-point Likert scale (1=strongly disagree, 7=strongly agree). Participants who answered five (somewhat agree) or higher could move to the next stage. Participants were randomly assigned to one of the two conditions which offered business or leisure travel scenarios. They were required to spend at least fifteen seconds reading the travel scenario. On the next page, their attitude about the hotel described in the scenario was measured. Participants were randomly assigned a second 31 time to one of the two conditions that differed in the message appeal (informational or emotional). As a result, the participants were exposed to one of the four conditions based on the combination of travel scenarios (business vs. leisure) and message appeal (informational vs. emotional). Participants were asked to imagine that they had searched Hotel XYZ on Instagram and read some of their posts. Then they were exposed to an Instagram post that had either informational or emotional appeal. They were required to spend at least fifteen seconds to read the Instagram post. Next, their attitude toward the hotel, attitude toward the post, engagement, and behavioral intention were measured, and manipulation checks were conducted. Manipulation checks For the travel purpose manipulation, the two manipulations questions were measured with a seven-point Likert scale. “The hotel stay is for a business trip.” (1=strongly disagree, 7=strongly agree), “The hotel stay is for a leisure purpose,” (1=strongly disagree, 7=strongly agree). The manipulation was successful for both the business trip (Mbusiness trip = 5.35, Mleisure trip = 1.20, df = 130, t = 3.741, p < .001) and leisure trip (Mbusiness trip = 4.20, Mleisure trip = 5.86, df = 130, t = -6.01, p < .001). For the appeal manipulation, the two items were measured: • The Instagram post was primarily factual. (1-strongly disagree to 7: strongly agree) • The Instagram post was primarily emotional (1-strongly disagree to 7: strongly agree) 32 From the pilot study, the manipulation results were shown to be successful with T- tests for both the emotional appeal manipulation check question (Minformational appeal = 4.19, Memotional appeal = 5.08, df = 130, t = -3.12 , p < .01) and the informational appeal manipulation check question (Minformational appeal = 5.42, Memotional appeal = 4.85, df = 130, t = 2.38 , p < .05). Thus, the Instagram posts were used for the main Study 1. Dependent variables Consumer engagement was measured for three dimensions (Table 4.1. Descriptive Means of Engagement – Study 1): interest and enjoyment (adapted from Mirbagheri & Najmi, 2019), page engagement (adapted from Gavilanes et al., 2018; Mirbagheri & Najmi, 2019;) and post engagement (adapted from Gavilanes et al., 2018). Participants were asked to indicate their agreement with the statements using a seven-point Likert scale ranging from strongly disagree to strongly agree. To measure behavioral intentions, three concepts were applied (Table 3.1): Search other source intentions (adapted from Sharifpour et al., 2014), word-of-mouth intention (excluding e-WOM; adapted from Chu & Kim, 2011; Harrison-Walker, 2001; Yang, 2013), and reservation intention (adopted from Leung & Jiang, 2018). Participants were asked to indicate their agreement with statements using a seven-point Likert scale ranging from strongly disagree to strongly agree. 33 Table 3. 1. Constructs and measure Engagement toward Social Media Message Interest and enjoyment • This Instagram post is playful. • Browsing Instagram posts related to this hotel is exciting. • Participating in this hotel's Instagram page is an enjoyable experience. This hotel's Instagram page is exciting Page engagement I will follow the posts related to this hotel (e.g., posts with the hashtag #HotelZEE). I will comment on the posts related to this hotel. I will share the posts related to this hotel. I will "like" the posts related to this hotel Post engagement I would be likely to click the "Like" button on this Instagram post. I would be likely to comment on this Instagram post. I would be likely to share this Instagram post. Behavioral Intentions Search other source intentions I would likely search the internet to learn more about this hotel. I would likely use other social media platforms to learn more about this hotel. I would likely use an online travel agency (e.g., Expedia.com) to learn more about this hotel. Word-of-mouth intention I would recommend this hotel to others. I would share information about this hotel with others in the future. I would say good things about this hotel. I would mention this hotel to others Buy intention I would reserve a room in this hotel. I would book this hotel room. I would stay in this hotel in the near future • • • • • • • • • • • • • • • 34 Study Participants and Data Cleaning Participants were recruited from the Amazon Mechanical Turk online survey system (MTurk, https://www.mturk.com) and directed to the online survey platform Qualtrics. Previous research confirmed that the data from online panels represent demographic diversity (Buhrmester, Kwang, & Gosling, 2011; Hwang et al., 2018; Jang & Mattila, 2018). Researchers (Jang & Mattila, 2018; Paolacci & Boegershausen, 2018) assert that these samples have benefits such as a reduction of experimenter expectancy effects or the elimination of social desirability bias. To eliminate the effect of repeated participation, those participants who participated in both pilot and main studies were removed from the data set of the main study, and additional participants were recruited for the main study. In Study 2, previous participants were excluded from the participants' pool using the Qualification function in MTurk (Paolacci & Boegershausen, 2018). In the pilot study of Study 1, each participant was rewarded with $0.20. For the main Study 1, each participant was rewarded with $1.00, and a total of $459.10, after fees and taxes, was paid to MTurk for 300 participants. From the first round of data collection, a total of 221 responses were collected. However, after reviewing the MTurk participants' data, 29 participants were found to have participated in the previous studies, including pretest and/or pilot study. Their responses were excluded. In the second round of data collection, an additional 147 participants were recruited for Study 1. Twenty-five participants failed the screening test (“I have experience with booking hotel rooms online”) on their online hotel room reservation (5: somewhat agree and higher passed). Among the questions measuring dependent variables, two attention check questions were embedded to ensure that participants paid attention to the survey. Forty-eight 35 responses were eliminated from the data analysis. Besides, 12 cases were identified as "careless respondents" based on highly repeated patterns in their responses. For example, if they answered most of the answers 7s or 1s, they were identified as careless respondents. These 12 data sets were excluded from the analysis. After excluding responses from repeat or careless participants and exclude responses based on screening questions, attention checks, in total, 254 responses were analyzed for Study 1. STUDY 2 Study 2 employed a 2 (information topic: core vs. supporting) X 2 (appeal: informational vs. emotional) between-subjects design, with subjects being randomly assigned to one of the four messages represented by different combinations of the two independent variables. This section presents the stimuli development process for supporting topic messages and the stimuli revision process for core topic messages from Study 1. Core topic - bedroom information Stimuli Development Researchers (Browning et al., 2013) recognized bedroom information as a core topic of hotel information. Study 1 tested the stimuli with a message containing bedroom features and room amenities information. For core topic messages of Study 2, the stimuli from Study 1 were adopted but revised. In Study 2, the core contents were strictly limited to room features such as cleanness, comfortableness, the spaciousness of the room, the view, and the bed. The social media message containing bedroom information was manipulated into two appeal messages (informational and emotional). 36 Supporting topic - concierge service information Service feature of hotel information was recognized as a non-core topic of hotel information (Dimitriadis & Koritos, 2014). To create a social media message featuring service information, multiple hotel social media messages were reviewed, and one social media message presenting the concierge service has been extracted and adapted. For appeal (informational or emotional) manipulation, similar rules and phrases to Study 1 were used. The social media message containing concierge service information was manipulated into two appeal messages (informational and emotional). Multiple pilot studies were conducted to test the manipulation for supporting topic messages, and the stimuli were revised to enhance the manipulation. Three pilot studies were conducted until the successful manipulation was observed. For Study 2, four Instagram messages were created (Appendix B). Assuming participants were not likely to be familiar with core and supporting services in a hotel context, a brief description of core service and supporting service was provided as follow: In a hotel context, there are two service aspects: core service vs. supporting service. Core Service represents a hotel's primary reason for being in the market. It comprises the hotel's fundamental competency in creating value with and for the customer. Examples of core services include providing a comfortable room and offering a clean bathroom. Supporting service creates ADDED value for the client and differentiates the firm's offering from those of its competitors. An example of supporting service is the concierge service. 37 Procedures (Pilot & Main Study) The procedures for Study 2 were similar to Study 1. After the consent form and screening test, the participants who passed the screening test were asked to imagine a general travel purpose scenario as below: You are going out of town on a trip for a couple of days. You will be staying at a four-star hotel (similar to the Holiday Inn, Four Seasons, Novotel, Radisson Hotel, and Suites) that offers semi high-end amenities such as personalized service, 24-hour room service, and valet parking, a fitness center, and a full-service restaurant. You found Hotel XYZ on the internet that matches your needs. Participants were required to spend at least fifteen seconds to read the scenario, and their attitude about the hotel described in the scenario was measured. In Study 2, participants were assigned to one of the four conditions based on the combination of message topics (core or supporting) and message appeal (informational or emotional). Manipulation checks Four manipulations check questions were asked to all the participants after the dependent variable questions were asked. After the manipulations, check questions, covariate questions, and demographic questions were followed. For the appeal manipulation, the two questions from Study 1 were asked as follows: • The Instagram post was primarily factual (1: strongly disagree to 7: strongly agree) • The Instagram post was primarily emotional (1: strongly disagree to 7: strongly agree) 38 From the last pilot study, the manipulation was successful using T-tests for both the emotional appeal manipulation check question (Minformational appeal = 4.30, Memotional appeal = 5.17, df = 277, t = -4.47, p < .001) and the informational appeal manipulation check question (Minformational appeal = 5.64, Memotional appeal = 4.96, df = 277, t = 4.18, p < .001). For the message topic manipulation, the two questions were asked: • The Instagram post described core services of the hotel (e.g., cleanliness), comfortableness and spaciousness of the room, the view, bed) (1: strongly disagree to 7: strongly agree) • The Instagram post described supporting services of the hotel (e.g., concierge service, spa service) (1: strongly disagree to 7: strongly agree) From the last pilot study, the manipulation was successful using T-tests for both the core topic manipulation check question (Mcore tooic = 5.53, Msupporting topic = 3.72, df = 277, t = -9.06, p < .001) and for supporting the topic manipulation check question (Mcore tooic = 4.81, |Msupporting topic = 6.04, df = 277, t = 7.40, p < .001). Study participants and data cleaning Before launching Study 2, the Amazon prototype was programmed to exclude previous participants from the study pool using Excel files. Over three hundred (n=322) Amazon MTurk workers participated, but twenty-three failed the screening test on their online hotel room reservation experience. Like in Study 1, two attention check questions (e.g., Please select strongly disagree) were planted, and thirty-four failed either of the two attention check questions, and their responses were deleted from the analysis. To detect any careless responses, including answering all strongly agree, the pattern of the responses was reviewed, but no extreme pattern was detected. In the end, 265 responses were analyzed. 39 In the first pilot study of Study 2, each participant was rewarded with $0.30, and a total of $66.50 for 160 participants was paid to MTurk, including fees. For the second pilot study, each participant was rewarded with $0.50 and a total of $112.00 after the fees were paid to MTurk for 160 participants. For the third pilot study, each participant was rewarded with $1.00, and a total of $420.00 for 300 participants was paid to MTurk, including fees. Finally, for the main study of Study 2, a total of $420 was paid for MTurk, including a reward of $1.00 per participant and a fee for 300 participants. 40 CHAPTER 4 RESULTS STUDY 1 In the following section, manipulation checks for the trip purpose (business vs. leisure) and message appeal (informational vs. emotional) are discussed to demonstrate how successfully the stimuli were manipulated. In step I of analysis, ANCOVA and ANOVA results are presented to investigate any interaction effects of trip purpose and message appeal on the three dimensions of engagement (i.e., interest and enjoyment, page engagement, post engagement). Testing of interaction effects is followed by planned contrasts to investigate how message appeal influences engagement depending on different trip scenarios (i.e., business trip vs. leisure trip). In step II of analysis, the results of a series of multiple regression analyses are presented to investigate whether levels of engagement can predict behavioral intention for the three dimensions (i.e., search intention, word-of-mouth intention, purchase intention). Research design Study 1 employed a 2 (trip purpose: business vs. leisure) × 2 (appeal: informational vs. emotional) between-subjects full-factorial design. The main purpose of Study 1 was to test the hypothesis that for a functional trip (i.e., business trip situation), an emotional appeal would generate higher engagement. Study 1 also tested the hypotheses that with the hedonic purpose of the trip (i.e., leisure trip situation), an informational appeal would generate higher engagement. Further, Study 1 tested the hypotheses that a higher level of engagement with the message (i.e., interest and enjoyment, page engagement, and post engagement) would predict 41 higher levels of behavioral intention (i.e., search intention, word-of-mouth intention, and purchase intention). The conceptual map of these hypotheses is presented (Figure 4.1). Figure 4. 1. Conceptual map– Study 1 Manipulation checks The experimental manipulations were successful. The results indicated that business trip scenario group recognized their purpose of trip as business trip (Mbusiness trip = 5.85, Mleisure trip = 3.88, df = 264, t = 7.94, p < .001), and leisure trip scenario group recognized their purpose of trip as leisure trip (Mbusiness trip = 4.42, Mleisure trip = 6.18, df = 264, t = - 8.03, p < .001). The results indicated that information appeal was perceived more informational than emotional (Minformational appeal = 5.54, Memotional appeal = 4.43 , df = 264, t = 6.45, p < .001), and emotional appeal was perceived as more emotional than informational (Minformational appeal = 4.61 , Memotional appeal = 5.42, df = 264, t = - 4.03, p < .001). Step I: The interaction effect of trip purpose and message appeal on engagement Dependent variables Interest and enjoyment, page engagement, and post engagement were measured as three dimensions of engagement. Descriptive means of three engagement dimensions (i.e., interest and enjoyment, page engagement, and post engagement) are displayed in Table 4.1. 42 Correlations between the three constructs of engagement were examined (Table 4.2). The correlation between interest and enjoyment and page engagement was significant (r = .71, df = 152, p < .001). The correlation between interest and enjoyment and post engagement was significant (r = .68, df = 152, p < .001). The correlation between page engagement and post engagement was significant (r = .93, df = 152, p < .001). Table 4. 1. Descriptive means of engagement – Study 1 Trip Purpose Business Leisure Appeal Informational Emotional n 122 132 n 131 123 Trip Purpose Appeal n Business Informational 63 Business Emotional 59 Leisure Informational 68 Leisure Emotional 64 Interest and enjoyment Engagement Page engagement Post engagement 5.09 (1.22) 5.09 (1.28) 4.48 (1.56) 4.41 (1.76) 4.48 (1.74) 4.40 (1.89) Interest and enjoyment Engagement Page engagement Post engagement 5.10 (1.21) 5.09 (1.30) 4.65 (1.54) 4.22 (1.76) 4.67 (1.72) 4.19 (1.89) Interest and enjoyment Engagement Page engagement Post engagement 4.96 (1.29) 5.22 (1.13) 5.22 (1.12) 4.95 (1.43) 4.40 (1.52) 4.56 (1.62) 4.88 (1.54) 3.91 (1.84) 4.48 (1.75) 4.47 (1.74) 4.84 (1.68) 3.93 (1.99) Note: Standard deviations are shown in parenthesis. Scale is 1: strongly disagree to 7: strongly agree. 43 Table 4. 2. Correlations, Means and Standard Deviations of Engagement– Study 1 Interest and enjoyment Page engagement Post engagement Mean Standard Deviation Interest and enjoyment 0.71 *** 0.68 *** 5.09 1.25 Page engagement Post engagement 0.93*** 4.44 1.66 4.43 Note: * p < .05, ** p < .01, *** p < .001, Scale is 1: strongly disagree to 7: strongly agree. A series of analyses of covariance (ANCOVA) were conducted to statistically control for the effect of covariates (Cronk, 2012). In the present study, one covariate, attitude toward the Instagram post, was identified and adopted as a significant factor influencing engagement with the message. The rest of the covariates tested were excluded from further analysis as they were not significant. Analysis of variance (ANOVA) was conducted to test the hypotheses and then followed by multiple regression analysis. Hypothesis 1: There will be the main effect of message appeal on engagement exhibited by (a) interest and enjoyment, (b) page engagement, and (c) post engagement. Hypothesis 2. Trip purpose moderates the effect of message appeal on the level of engagement. More specifically H2a: In the functional (vs. hedonic) trip scenario, the level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will be higher with emotional appeal than with informational appeal. H2a(a). In the functional trip scenario, the level of interest and enjoyment with the message will be higher with emotional appeal than with informational appeal. H2a(b). In the functional trip scenario, the level of engagement page engagement with the message will be higher with emotional appeal than with informational appeal. H2a(c). In the functional trip scenario, the level of engagement (post engagement with the message will be higher with emotional appeal than with informational appeal. 44 H2b. In the hedonic (vs. functional) trip scenario, the level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will be higher in informational appeal than in emotional appeal. H2b(a). In the hedonic trip scenario, the level of interest and enjoyment with the message will be higher in informational appeal than in emotional appeal. H2b(b). In the hedonic trip scenario, the level of page engagement with the message will be higher in informational appeal than in emotional appeal. H2b(c). In the hedonic trip scenario, the level of post engagement with the message will be higher in informational appeal than in emotional appeal. Hypothesis 3. The level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will positively influence the behavioral intention (a. search intention, b. word-of-mouth intention, c. purchase intention) H3a. The level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will positively influence the search intention. H3a(a). The level of interest and enjoyment with the message will positively influence the search intention. H3a(b). The level of page engagement with the message will positively influence the search intention. H3a(c). The level of post engagement with the message will positively influence the search intention. H3b. The level of engagement ((a) interest and enjoyment, (b) page engagement, (c) post engagement) with the message will predict the level of word-of-mouth intention. H3b(a). The level of interest and enjoyment with the message will positively influence the word-of-mouth intention. H3b(b). The level of page engagement with the message will positively influence word-of-mouth intention. H3b(c). The level of post engagement with the message will positively influence the word-of-mouth intention. H3c. The level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will predict the level of purchase intention. H3c(a). The level of interest and enjoyment with the message will predict the level of purchase intention. H3c(b). The level of page engagement with the message will predict the level of purchase intention. H3c(c). The level of post engagement with the message will predict the level of purchase intention. 45 Interest and enjoyment A 2 (trip purpose: business vs. leisure) × 2 (appeal: informational vs. emotional) factorial between-subjects ANCOVA was estimated to examine the effect of trip purpose on interest and enjoyment, controlling for any effect of attitude toward the Instagram post. Attitude toward the Instagram post was significantly related to interest and enjoyment (F(1, 249) = 223.48, p < .001)(Table 4.3). The main effect of trip purpose on interest and enjoyment was not significant (F(1, 249) = 1.26, p = .264) with the business trip purpose group (M = 5.09, sd = 1.22 ) not showing significantly different level of interest and enjoyment from leisure trip group (M = 5.09, sd = 1.28). The main effect of appeal on interest and enjoyment was not significant (F(1, 249) = 2.76, p = .10 with the informational appeal group (M = 5.10, sd = 1.21) not showing significantly different level of interest and enjoyment from the emotional appeal group (M = 5.09, sd = 1.30). The ANCOVA results indicated that the interaction effect of trip purpose and appeal on interest and enjoyment was significant (F(1,249) = 6.83, p < .05). The interaction effect is visualized in Figure 4.1. Planned contrasts showed no significant effect of appeal on interest and enjoyment in both business trip (Minformational appeal = 4.96, Memotional appeal = 5.22, t(120) = -1.21 , p = .23) and leisure trip (Minformational appeal = 5.22, Memotional appeal = 4.95, t(130) = 1.25 , p = .21) groups. 46 Table 4. 3. Means and ANCOVA results: The impacts of trip and appeal on interest and enjoyment Means Leisure Mean 5.10 5.09 5.10 Appeal Informational Emotional Mean 4.96 5.22 5.09 5.22 4.95 5.09 Trip Purpose Business Analysis of covariance Type III Sum of Squares Mean Square 184.32 223.48 *** 1 F df 184.32 Source Covariate Attitude toward the Instagram post Test Effects Trip Appeal Trip * Appeal Error Total Corrected Total Note. R2 = .48 (Adjusted R2 = .47), * = p < .05, ** = p < .01, *** = p < .001 1.03 2.28 5.63 205.37 6971.38 394.38 1.03 2.28 5.63 .83 1 1 1 249 254 253 1.26 2.76 6.83 * P-value .000 .264 .098 .010 Figure 4. 2. The interaction effect of trip and appeal on interest and enjoyment j t n e m y o n E d n a t s e r e t n I y l g n o r t s : 7 , e e r g a s i d y l g n o r t s : 1 ( . ) 6 3 t a d e t a c n u r t , e e r g a 5.4 5.2 5.0 4.8 4.6 4.4 4.2 4.0 3.8 3.6 Informational Appeal Emotional Appeal Business Trip Leisure Trip 47 Without controlling for the effect of attitude toward the Instagram post, the ANOVA results (Table 4.4) showed that the interaction effect of trip purpose and message appeal on interest and enjoyment was not significant (F(1,250) = 4.67, p = .09). On the other hand, when the attitude toward the Instagram post was controlled, the ANCOVA result showed a significant interaction effect, supporting the significant role of attitude toward the Instagram post and justifying ANCOVA in this study. Table 4. 4. ANOVA results: The impacts of trip and appeal on interest and enjoyment Type III Sum of Squares Source Test Effects Trip Appeal Trip * Appeal Error Total Corrected Total Note. R2 = .01 (Adjusted R2 = .000), * = p < .05, ** = p < .01, *** = p < .001 .01 .01 4.67 389.69 6971.38 394.38 df 1 1 1 250 254 253 Mean Square .00 .00 4.67 1.56 F .00 .002 3.00 p-value .96 .96 .085 Page engagement A 2 (trip purpose: business vs. leisure) × 2 (appeal: informational vs. emotional) factorial between-subjects ANCOVA (Table 4.4) was calculated to examine the effect of trip purpose on page engagement, covarying out the effect of attitude toward the Instagram post. Attitude toward the Instagram post was significantly related to page engagement (F(1, 249) = 77.99, p < .001) (Table 4.4). The main effect of trip purpose on page engagement was not significant (F(1, 249) = .05, p = .823) with the business trip group ( M = 4.48, sd =1.56) showed no significantly higher level of participation than the leisure trip group (M = 4.41, sd =1.76 ). The main effect of appeal on page engagement was not significant 48 (F(1, 249) = 1.49, p = .223) with the informational appeal group ( M = 4.65, sd =1.54) showed no significantly higher level of page engagement than the emotional appeal group (M = 4.22, sd =1.76). The ANCOVA results (Table 4.5) indicated that the interaction effect of trip purpose and message appeal on page engagement was significant (F(1,249) = 11.04, p < .01). The interaction effect is visualized in Figure 4.3. Planned contrasts showed that when the trip purpose is leisure, informational appeal compared to emotional appeal caused higher levels of page engagement (Minformational appeal = 4.88, Memotional appeal = 3.91, t(130) = 3.31 , p < .01). However, among the business trip group, no significant effect of appeal on page engagement was found (Minformational trip = 4.40, Memotional appeal = 4.56, t(120) = -.59 , p =.56). 49 Table 4. 5. Means and ANCOVA results: The impacts of trip and appeal on page engagement Means Analysis of covariance Trip Purpose Appeal Business Leisure Mean 4.65 4.22 Informational Emotional Mean 4.40 4.56 4.48 4.88 3.91 4.41 Type III Sum of Squares Mean Square df F 1 159.01 159.01 Source Covariate Attitude toward the Instagram post Test Effects Trip Appeal Trip * Appeal Error Total Corrected Total Note. R2 =.27 (Adjusted R2 =.26), * = p < .05, ** = p < .01, *** = p < .001 .10 3.04 22.50 507.64 5710.81 699.21 .10 3.04 22.50 2.04 1 1 1 249 254 253 .05*** 1.49*** 11.04*** 77.99*** P- value .000 .823 .223 .001 Figure 4. 3. The interaction effect of trip and appeal on page engagement : 7 , e e r g a s i d y l g n o r t s : 1 ( t n e m e g a g n E e g a P t a d e t a c n u r t , e e r g a y l g n o r t s ) 6 3 . 5.4 5.2 5.0 4.8 4.6 4.4 4.2 4.0 3.8 3.6 Informational Appeal Emotional Appeal Business Trip Leisure Trip The ANOVA results (Table 4.6) indicated that the interaction effect of trip purpose and message appeal on page engagement was significant (F(1, 250) = 20.68, p < .01). 50 Table 4. 6. ANOVA results: The impacts of trip and appeal on page engagement Source Test Effects Trip Appeal Trip * Appeal Error Total Corrected Total Note. R2 = .05 (Adjusted R2 = .04), * = p < .05, ** = p < .01, *** = p < .001 .467 10.37 20.68 2.67 Type III Sum of Squares .47 10.37 20.68 666.65 5710.81 699.21 Mean Square df 1 1 1 250 254 253 F .18 3.89 * 7.76 ** Post engagement A 2 (trip purpose: business vs. leisure) × 2 (appeal: informational vs. emotional) factorial between-subjects ANCOVA (Table 4.7) was estimated to examine the effect of trip purpose on post engagement, covarying out the effect of attitude toward the Instagram post. Attitude toward the Instagram post was significantly related to post engagement (F(1, 249) = 152.58, p < .001) (Table 4.7). The main effect of trip purpose on post engagement was not significant (F(1, 249) = .06, p = .879) with the business trip group (M = 4.48, a sd =1.74) with no significantly higher level of post engagement than the leisure trip group (M = 4.40, sd =1.89 ). The main effect of appeal on post engagement was not significant (F(1, 249) = 4.95, p = .170) with the informational appeal group (M = 4.67, sd =1.72) with no significantly higher level of post engagement than the emotional appeal group (M = 4.19, sd =1.89). The ANCOVA results indicated that the interaction effect of trip purpose and message appeal on post engagement was significant (F(1, 249)= 5.42, p < .05). The interaction effect is visualized in Figure 4.3. Planned contrasts showed that when the trip purpose is leisure, informational appeal compared to emotional appeal caused higher levels of post engagement (Minformational appeal = 4.84, Memotional appeal = 3.93, t(130)= 2.85 , p < .01). 51 However, among the business trip group, no significant effect of appeal on post engagement (Minformational appeal = 4.48, Memotional appeal = 4.47, t(120) = .04 , p = .97) was observed. Trip Purpose Table 4. 7. Means and ANCOVA results: The impacts of trip and appeal on post engagement Means Analysis of covariance Appeal Informational Emotional Mean Leisure Mean 4.67 4.19 4.84 3.93 4.40 4.48 4.47 4.48 Business Type III Sum of Squares Mean Square 152.58 58.31 *** P- value 1 F df 152.58 Source Covariates Attitude toward the Instagram post Test Effects Trip Appeal Trip * Appeal Error Total Corrected Total Note. R2 = .22 (Adjusted R2 = .20), * = p < .05, ** = p < .01, *** = p < .001 .06 4.95 14.19 651.61 5826.56 831.96 .06 4.95 14.19 2.62 1 1 1 249 254 253 .02 1.89 5.42 * .000 .879 .170 .021 52 Figure 4. 4. The interaction effect of trip and appeal on post engagement t n e m e g a g n E t s o P , e e r g a s i d y l g n o r t s : 1 ( d e t a c n u r t , e e r g a y l g n o r t s : 7 . ) 6 3 t a 5.4 5.2 5.0 4.8 4.6 4.4 4.2 4.0 3.8 3.6 Business Trip Leisure Trip Informational Appeal Emotional Appeal The ANOVA results (Table 4.8) indicated that the interaction effect of trip purpose and message appeal on post engagement was significant (F(1, 250) = 12.79, p < .05). Table 4. 8. ANOVA results: The impacts of trip and appeal on post engagement Source Test Effects Trip Appeal Trip * Appeal Error Total Corrected Total Note. R2 = .03 (Adjusted R2 = .02), * = p < .05, ** = p < .01, *** = p < .001 .54 13.51 12.79 3.22 Type III Sum of Squares .54 13.51 12.79 804.19 5826.56 831.96 Mean Square df 1 1 1 250 254 253 F .17 4.20 * 3.98 * Step II: Engagement as a predictor of behavioral intention Dependent variables A series of multiple regression analyses were conducted to examine the effects of engagement types or levels on behavioral intentions. Search intention, word-of-mouth intention, and purchase intention were measured and tested as three dimensions of behavioral 53 intention. Descriptive means of three dimensions of behavioral intention are shown in Table 4.9. Table 4. 9. Descriptive Means of Behavioral Intention – Study 1 Trip Purpose Business Leisure n 122 132 Behavioral Intention Search intention Word-of- mouth intention Purchase intention 5.54 (0.97) 5.22 (1.34) 5.16 (1.14) 5.12 (1.23) 5.39 (1.02) 5.37 (1.24) Note: Standard deviations are shown in parenthesis. Scale is 1: strongly disagree to 7: strongly agree. Appeal Informational Emotional n 131 123 Behavioral Intention Search intention Word-of- mouth intention Purchase intention 5.55 (0.97) 5.18 (1.36) 5.23 (1.15) 5.04 (1.22) 5.44 (1.11) 5.31 (1.17) Note: Standard deviations are shown in parenthesis. Scale is 1: strongly disagree to 7: strongly agree. Trip Purpose Business Appeal Informational Business Emotional Leisure Informational Leisure Emotional n 63 59 68 64 Behavioral Intention Search intention Word-of- mouth intention Purchase intention 5.57 (0.99) 5.51 (0.97) 5.53 (0.96) 4.88 (1.60) 5.15 (1.18) 5.18 (1.11) 5.32 (1.12) 4.91 (1.32) 5.48 (0.98) 5.28 (1.07) 5.42 (1.22) 5.33 (1.14) Note: Standard deviations are shown in parenthesis. Scale is 1: strongly disagree to 7: strongly agree. 54 Table 4. 10. Correlations, Means and Standard Deviations of Behavioral Intention Search intention 0.55 *** Word-of-mouth intention Purchase intention Search intention Word-of-mouth intention Purchase intention Mean Standard Deviation 5.38 1.14 Note: * p < .05, ** p < .01, *** p < .001, Scale is 1: strongly disagree to 7: strongly agree. 0.70 *** 5.14 1.19 0.58 *** 5.37 1.19 Search intention Hypothesis 3a suggested that interest and enjoyment, page engagement, and post engagement would predict search intention. Multiple regression analysis was used to analyze the relationships among the variables. The result was significant (F(3, 250) = 39.87, p < .001, adj. R 2= .32). The analysis showed that interest and enjoyment (β = .46, t = 6.17, p < .001) was a significant predictor of search intention, while page engagement (β = .203, t = 1.340, p = .181) and post engagement (β = - .07, t = - .45, p = .66) were not significant. Thus, it is concluded that more interest and enjoyment is associated with more search intention supporting H3a(a). Word-of-mouth intention Hypothesis 3b suggested that interest and enjoyment, page engagement, and post engagement would predict word-of-mouth intention. Multiple regression analysis was used to analyze the relationships among the variables. The result was significant (F(3, 250) = 141.18, p < .001, adj. R 2= .62). The analysis showed that interest and enjoyment (β = .42, t = 7.57, p < .001) and page engagement (β = .43, t = 3.80, p < .001) were significant predictors of word-of-mouth intention, while post engagement (β = .01, t = .13, p = .90) was not. Thus, it 55 was concluded that more interest and enjoyment and more page engagement caused more word-of-mouth intention, thus supporting H3b(a) and H3b(b). Purchase intention Hypothesis 3c suggested that interest and enjoyment, page engagement, and post engagement would predict purchase intention. Multiple regression analysis was used to analyze the relationships among the variables. The result was significant, F (3, 250) = 75.42, p < .001, adj. R 2= .47 The analysis showed that interest and enjoyment (β = .70, t = 10.64, p < .001) was a significant predictor of purchase intention, while page engagement (β = .09, t = .65, p = .52) and post engagement (β = - .10 t = - .801, p = .42) were not significant. Thus, it was concluded that more interest and enjoyment indicated more purchase intention supporting H3c(a). STUDY 2 In the following section, manipulation checks for message topic (core vs. supporting) and message appeal (informational vs. emotional) are presented to illustrate how successful the stimuli were manipulated. The ANCOVA and ANOVA results are displayed in the following section to investigate the interaction effects of message topic and message appeal on three dimensions of engagement (interest and enjoyment, page engagement, post engagement). As no significant interaction effect of message topic and appeal on engagement was observed in Study 2, the main effects of message topic and message appeal on engagement are discussed. 56 Research design Study 2 employed a 2 (message topic: core vs. supporting) × 2 (appeal: informational vs. emotional) between-subjects design. Participants were randomly assigned to one of the four conditions. The primary purpose of Study 2 was to test the hypothesis that with core information of the hotel (i.e., room feature information), the informational appeal would generate a higher level of engagement. Study 2 tested the hypothesis that with the hotel's supporting information (i.e., service feature information), the emotional appeal would generate a higher engagement level. An Instagram post with core hotel attributes (i.e., bedroom information) was assumed to create the central route in this study, as researchers support that room feature is a core attribute in a hotel context and is associated with high involvement. Figure 4.5 illustrates these hypotheses. Figure 4. 5. Conceptual map – Study 2 Manipulation checks The experimental manipulations were successful. The results indicated that core information group recognized the topic of the Instagram post as core information (Mcore information = 5.71, Msupporting information = -8.74, t(263) = - 8.61, p < .001), and supporting information group recognized the topic of the Instagram post as supporting information (Mcore information = 5.27, Msupporting information = 6.12 , t(263)= 5.36, p < .001). The results indicated 57 that information appeal group recognized their Instagram post as informational (Minformational trip = 5.89, Memotional appeal = 5.28, t(263)= 4.46, p < .001), and emotional appeal group recognized their Instagram post as emotional (Minformational trip = 4.72, Memotional appeal = 5.41, t(263) = -3.65, p < .001). Step I: The interaction effect of message topic and message appeal on engagement Dependent variables Interest and enjoyment, page engagement, and post engagement were measured as three dimensions of engagement. Descriptive means of three engagement dimensions are shown in Table 4.11. Correlations between the three constructs of engagement were examined (Table 4.12). The correlation between interest and enjoyment and page engagement was significant (r = .74, df = 263, p < .001). The correlation between interest and enjoyment and post engagement was significant (r = .73, df = 263, p < .001). The correlation between page engagement and post engagement was significant (r = .95, df = 263, p < .001). A series of ANCOVAs and ANOVAs were conducted to test the hypotheses. To be consistent with Study 1, attitude toward the Instagram post was included in the ANCOVA analysis as a covariate. 58 Table 4. 11. Descriptive Means of Engagement – Study 2 Message Topic Supporting Core Appeal Informational Emotional Message Topic Supporting Appeal Informational Supporting Emotional Core Core Informational Emotional Interest and enjoyment Engagement Page engagement Post engagement 5.30 (1.23) 5.20 (1.11) 4.74 (1.67) 4.57 (1.66) 4.73 (1.72) 4.67 (1.74) Interest and enjoyment Engagement Page engagement Post engagement 5.04 (1.34) 5.46 (0.93) 4.51 (1.78) 4.79 (1.53) 4.53 (1.81) 4.87 (1.64) Interest and enjoyment Engagement Page engagement Post engagement 5.11 (1.46) 5.48 (0.92) 4.97 (1.22) 5.42 (0.93) 4.64 (1.87) 4.83 (1.46) 4.39 (1.69) 4.76 (1.61) 4.59 (1.88) 4.87 (1.54) 4.47 (1.74) 4.87 (1.73) n 129 136 n 133 132 n 64 65 69 67 Note: Standard deviations are shown in parenthesis, Scale is 1: strongly disagree to 7: strongly agree Table 4. 12. Correlations, Means and Standard Deviations of Engagement– Study 2 Interest and enjoyment Page engagement Post engagement Mean Standard Deviation Interest and enjoyment 0.74 *** 0.73 *** 5.24 1.17 Page engagement Post engagement 0.95 *** 4.65 1.66 4.70 1.73 Note: * p < .05, ** p < .01, *** p < .001, Scale is 1: strongly disagree to 7: strongly agree. A series of ANCOVAs have been conducted covarying out attitude toward the Instagram post to test the hypotheses. A series of analyses of variances (ANOVAs) were also conducted to test the hypotheses below. 59 Hypothesis 1: There will be the main effect of message appeal on engagement exhibited by (a) interest and enjoyment, (b) page engagement, (c) post engagement. Hypothesis 2. Message topic moderates the effect of message appeal on the level of engagement. More specifically H2a: With core (vs. supporting) information, the level of engagement ((a) interest and enjoyment, (b) page engagement, (c) post engagement) with the message will be higher with emotional appeal than with informational appeal. H2a(a). With core (vs. supporting) information, the level of interest and enjoyment with the message will be higher with emotional appeal than with informational appeal. H2a(b). With core (vs. supporting) information, the level of engagement page engagement with the message will be higher with emotional appeal than with informational appeal. H2a(c). With core (vs. supporting) information, the level of engagement (post engagement with the message will be higher with emotional appeal than with informational appeal. H2b. With supporting (vs. core) information, the level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will be higher in informational appeal than in emotional appeal. H2b(a). With supporting (vs. core) information, the level of interest and enjoyment with the message will be higher in informational appeal than in emotional appeal. H2b(b). With supporting (vs. core) information, the level of page engagement with the message will be higher in informational appeal than in emotional appeal. H2b(c). With supporting (vs. core) information, the level of post engagement with the message will be higher in informational appeal than in emotional appeal. Hypothesis 3. The level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will positively influence the behavioral intention (a. search intention, b. word-of-mouth intention, c. purchase intention) H3a. The level of engagement ((a) interest and enjoyment, (b) page engagement, and (c) post engagement) with the message will positively influence the search intention. H3a(a). The level of interest and enjoyment with the message will positively influence the search intention. H3a(b). The level of page engagement with the message will positively influence the search intention. 60 H3a(c). The level of post engagement with the message will positively influence the search intention. H3b. The level of engagement ((a) interest and enjoyment, (b) page engagement, (c) post engagement) with the message will predict the level of word-of-mouth intention. H3b(a). The level of interest and enjoyment with the message will positively influence the word-of-mouth intention. H3b(b). The level of page engagement with the message will positively influence word-of-mouth intention. H3b(c). The level of post engagement with the message will positively influence the word-of-mouth intention. H3c. The level of engagement ((a) interest and enjoyment, (b) page engagement, (c) post engagement) with the message will predict the level of purchase intention. H3c(a). The level of interest and enjoyment with the message will predict the level of purchase intention. Interest and enjoyment A 2 (message topic: core vs. supporting) × 2 (appeal: informational vs. emotional) factorial between-subjects ANCOVA (Table 4.13) was calculated to examine the effect of the message topic on interest and enjoyment covarying out the effect of attitude toward the Instagram post. Attitude toward the Instagram post was significantly related to interest and enjoyment (F(1, 260) = 115.89, p < .001). The main effect of message topic on interest and enjoyment was not significant (F(1, 260) = .050, p = .824) with core information group (M = 5.20, sd = 1.11) showing insignificantly higher level of interest and enjoyment from supporting information group (M = 5.09, sd = 1.28). The main effect of appeal on interest and enjoyment was significant (F(1, 260) = 10.25, p < .01) with emotional appeal group (M = 5.46, sd = 0.93) showing a significantly higher level of interest and enjoyment than informational appeal group (M = 5.04, sd = 1.34). The ANCOVA results indicated that the 61 interaction effect of topic and appeal on interest and enjoyment was not significant (F(1,260) = .34, p = .559). The effect is visualized in Figure 4.6. Table 4. 13. Means and ANCOVA results: The impacts of topic and appeal on interest and enjoyment Means Supporting Mean 5.04 5.46 Appeal Informational Emotional Mean Core 4.97 5.42 5.20 5.11 5.48 5.30 Message Topic Analysis of covariance Type III Sum of Squares Mean Square 107.25 F 115.89 *** 1 df 107.25 Source Covariate Attitude toward the Instagram post Test Effects Topic Appeal Topic * Appeal Error Total Corrected Total Note. R2 = .33 (Adjusted R2 = .32), * = p < .05, ** = p < .01, *** = p < .001 .05 9.49 .32 240.63 7643.19 360.11 1 1 1 260 265 264 .05 9.49 .32 .34 .050 10.25 ** P-value .000 .824 .002 .559 62 Figure 4. 6. The effect of topic and appeal on interest and enjoyment j t n e m y o n E d n a t s e r e t n I y l g n o r t s : 7 , e e r g a s i d y l g n o r t s : 1 ( . ) 6 3 t a d e t a c n u r t , e e r g a 5.6 5.4 5.2 5.0 4.8 4.6 4.4 4.2 4.0 3.8 3.6 Informational Appeal Emotional Appeal Room Information Service Information The ANOVA results (Table 4.14) indicated that the interaction effect of topic and appeal on interest and enjoyment was not significant (F(1,261)=1.33, p= .791). Table 4. 14. ANOVA results: The impacts of topic and appeal on interest and enjoyment Type III Sum of Squares df Source Test Effects Topic Appeal Topic * Appeal Error Total Corrected Total Note. R2 = .03 (Adjusted R2 = .02), * = p < .05, ** = p < .01, *** = p < .001 .72 11.28 .09 347.88 7643.19 360.11 1 1 1 261 265 264 Mean Square .72 11.28 .09 1.33 8.47** F .54 .07 P-value .463 .004 .791 Page engagement A 2 (message topic: core vs. supporting) × 2 (appeal: informational vs. emotional) factorial between-subjects ANCOVA (Table 4.15) was calculated to examine the effect of 63 message topic on page engagement, covarying out the effect of attitude toward the Instagram post. Attitude toward the Instagram post was significantly related to page engagement (F(1, 260) = 22.58, p < .001). The main effect of message topic on page engagement was not significant (F(1, 260) = .29, p = .590) with core information group (M = 4.57, sd = 1.66) showing insignificantly lower level of page engagement from supporting information group (M = 4.74, sd = 1.67). The main effect of appeal on page engagement was not significant (F(1, 260) = 1.63, p = .202) with informational appeal group (M = 4.51, sd = 1.78) showing insignificantly lower level of page engagement than emotional appeal group (M = 4.79, sd = 1.53).The ANCOVA results indicated that the interaction effect of topic and appeal on page engagement was not significant (F(1,260) = .35, p = .55). The effect is visualized in Figure 4-7. 64 Table 4. 15. Means and ANCOVA results: The impacts of topic and appeal on page engagement Means Analysis of covariance Supporting Mean 4.51 4.79 Appeal Informational Emotional Mean Core 4.39 4.76 4.57 4.64 4.83 4.74 Message Topic Type III Sum of Squares Mean Square 22.58 *** 1 F df 57.60 57.60 Source Covariates Attitude toward the Instagram post Test Effects Topic Appeal Topic * Appeal Error Total Corrected Total Note. R2 = .09 (Adjusted R2 = .08), * = p < .05, ** = p < .01, *** = p < .001 .74 4.17 .90 663.31 6460.75 728.46 .74 4.17 .90 2.55 1 1 1 260 265 264 .29 1.63 .35 P-value .000 .590 .202 .552 Figure 4. 7. The effect of topic and appeal on page engagement . ) 6 3 t a d e t a c n u r t , e e r g a t n e m e g a g n E e g a P y l g n o r t s : 7 , e e r g a s i d y l g n o r t s : 1 ( 5.4 5.2 5 4.8 4.6 4.4 4.2 4 3.8 3.6 Informational Appeal Emotional Appeal Room Information Service Information 65 The ANOVA results (Table 4.16) indicated that the interaction effect of topic and appeal on interest and enjoyment was not significant (F(1,261) = .21, p = .647). Table 4. 16. ANOVA results: The impacts of topic and appeal on page engagement Type III Sum of Squares 1.76 5.05 .58 720.90 6460.75 728.46 df 1 1 1 261 265 264 Mean Square 1.76 5.05 .58 2.76 F .64 1.83 .21 P-value .425 .178 .647 Source Test Effects Topic Appeal Topic * Appeal Error Total Corrected Total Note. R2 = .01 (Adjusted R2 = -.00) Post engagement A 2 (message topic: core vs. supporting) × 2 (appeal: informational vs. emotional) factorial between-subjects ANCOVA (Table 4.17) was calculated to examine the effect of message topic on post engagement, covarying out the effect of attitude toward the Instagram post. Attitude toward the Instagram post was significantly related to post engagement (F(1, 260) = 21.55, p < .001). The main effect of message topic on post engagement was not significant (F(1, 260) = .00, p = . 99) with core information group (M = 4.67, sd = 1.74) showing insignificantly lower level of post engagement from supporting information group (M = 4.73, sd = 1.72). The main effect of appeal on post engagement was not significant (F(1, 260) = .15, p = .703) with the informational appeal group (M = 4.53, sd = 1.81) showing insignificantly lower levels of post engagement than the emotional appeal group (M = 4.87, sd = 1.64). The ANCOVA results indicated that the interaction effect of topic and 66 appeal on post engagement was not significant (F(1,260) = .40, p = .15). The effect is visualized in Figure 4.8. Table 4. 17. Means and ANCOVA results: The impacts of topic and appeal on post engagement Means Analysis of covariance Supporting Mean 4.53 4.87 Appeal Informational Emotional Mean Core 4.47 4.87 4.67 4.59 4.87 4.73 Message Topic Type III Sum of Mean Square F 59.69 21.55 *** P- value 1 df 59.69 Squares Source Covariates Attitude toward the Instagram post Test Effects Topic Appeal Topic * Appeal Error Total Corrected Total Note. R2 = .09 (Adjusted R2 = .07), * = p < .05, ** = p < .01, *** = p < .00 .00 6.45 .40 720.28 6637.22 788.07 .00 6.45 .40 2.77 1 1 1 260 265 264 .00 2.33 .15 .000 .990 .128 .703 67 Figure 4. 8. The interaction effect of topic and appeal on post engagement 5.4 5.2 5 4.8 4.6 4.4 4.2 4 3.8 3.6 Informational Appeal Emotional Appeal y l g n o r t s : 7 , e e r g a s i d y l g n o r t s : 1 ( t n e m e g a g n E t s o P . ) 6 3 t a d e t a c n u r t , e e r g a Room Information Service Information The ANOVA results (Table 4.18) indicated that the interaction effect of topic and appeal on post engagement was not significant (F(1,261) = 2.13, p = .798). Table 4. 18. ANOVA results: The impacts of topic and appeal on post engagement Source Test Effects Topic Appeal Topic * Appeal Error Total Corrected Total Note. R2 = .02 (Adjusted R2 = .01) Type III Sum of Squares 2.21 .07 1.86 228.10 8272.33 232.23 df 1 1 1 261 265 264 Mean Square 2.21 .07 1.86 .87 F P-value 2.53 .08 2.13 .776 .113 .798 Unlike Study 1, no multiple regression analysis was conducted, as no significant effects of topic and appeal on engagement were observed. 68 SUMMARY OF RESULTS The main purpose of Study 1 and Study 2 was to test the hypothesis that trip purpose (Study 1) and message topic (Study 2) moderate the effect of message appeal on the level of engagement (H2). The ANCOVA results (Table 4.19) from Study 1 supported this hypothesis (H2) with significant interaction effects of trip purpose and message appeal on engagement across all three dimensions of engagement (interest and enjoyment, page engagement, post engagement). However, the ANCOVA results from Study 2 rejected this hypothesis (H2) with no significant interaction effects of message topic and message appeal on engagement. Table 4. 19. Univariate F-values for engagement (interest and enjoyment, page engagement, post engagement) Study 1 Trip Appeal Trip * Appeal Study 2 Topic Appeal Topic * Appeal Interest and enjoyment Page engagement Post engagement 1.26 2.76 6.83* .050 10.25** .34 .05 1.49 11.04** .29 1.63 .35 .02 1.89 5.42* .00 2.33 .15 * = p < .05, ** = p < .01, *** = p < .001 Study 1 and Study 2 also tested the main effect of message appeal (informational vs. emotional) on engagement (H1) (Table 4.16). Study 2 partially supported the main effect of appeal on engagement (H1) with a higher level of interest and enjoyment in the emotional appeal group than in the informational appeal group. Study 1 did not support the main effect of message appeal on engagement (H1). Multiple regression results partially supported the hypothesis that the level of engagement with the message predicts the level of word-of-mouth. Intention (H3) (Table 4.17a). First, more interest and enjoyment indicated more search intention (H3a(a)). More interest and enjoyment, and more page engagement indicated more word-of-mouth intention 69 (H3b(a), H3b(b)). At last, more interest and enjoyment indicated more purchase intention (H3c(a)). The summary of hypothesis testing results for the main effect and regression analysis have been presented in Table 4.20. Table 4. 20. Summary of the results of hypothesis testing a. Main effect and regression analysis IVs Appeal (study1) Appeal (Study 2) Interest and enjoyment Page engagement Post engagement DVs H1(a) Interest and enjoyment H1(b) Page engagement H1(c) Post engagement H1(a) Interest and enjoyment H1(b) Page engagement H1(c) Post engagement H3a(a) Search intention H3b(a) Word-of-mouth intention H3c(a) Purchase intention H3a(b) Search intention H3b(b) Word-of-mouth intention H3c(b) Purchase intention H3a(c) Search intention H3b(c) Word-of-mouth intention H3c(c) Purchase intention Result Not supported Not supported Not supported Supported Not supported Not supported Supported Supported Supported Not supported Supported Not supported Not supported Not supported Not supported b. Interaction effect – Study 1 Study1 IVs Business Trip Leisure Trip Informational Emotional Results Interest and enjoyment (a) H2a(a) 4.96 (1.29) !!! 4.95 (1.43) Not Supported Page engagement (b) Business Trip H2a(b) 4.40 (1.52) ** 3.91 (1.84) Supported Post engagement (c) H2a(c) 4.48(1.75) H2b(c) 4.84 (1.68) ≈!!!! 4.47 (1.74) Not Supported >** 3.93(1.99) Supported Note: Standard deviations are shown in parenthesis. Scale is 1: strongly disagree to 7: strongly agree. Planned contrasts: *p < .05. **p <.01 70 Table 4. 20. (cont’d) c. Interaction effect – Study 2 Study 2 IVs Core Topic Supporting Topic Core Topic Supporting Topic Core Topic Supporting Topic H2a(a) H2b(a) H2a(b) H2b(b) H2a(c) H2b(c) Informational Emotional Results Interest and enjoyment (a) 4.97(1.22) 5.11(1.46) <*